diff --git a/studio/backend/auth/authentication.py b/studio/backend/auth/authentication.py
index 2481cd13e6..94df994928 100644
--- a/studio/backend/auth/authentication.py
+++ b/studio/backend/auth/authentication.py
@@ -164,14 +164,13 @@ async def get_current_subject_allow_password_change(
     )
 
 
-# The literal the examples ship with; pasting one unedited is likelier than a revoked key.
+# The literal the examples ship with; pasted unedited more often than a revoked key.
 API_KEY_PLACEHOLDER = f"{API_KEY_PREFIX}YOUR_KEY"
 
 
 def _invalid_api_key_detail(token: str) -> str:
-    """Why the key failed. Only the unedited example placeholder is called out;
-    every real key still gets one indistinguishable message, so this reveals
-    nothing about which keys exist."""
+    """Why the key failed. Only the example placeholder is called out; every real
+    key gets one indistinguishable message, so this leaks no key existence."""
     if token == API_KEY_PLACEHOLDER:
         return (
             "This is the placeholder key from the example. Create an API key in "
diff --git a/studio/backend/core/inference/api_monitor.py b/studio/backend/core/inference/api_monitor.py
index 2de042ab37..ce32a6d3ef 100644
--- a/studio/backend/core/inference/api_monitor.py
+++ b/studio/backend/core/inference/api_monitor.py
@@ -148,10 +148,9 @@ class ApiMonitor:
     ) -> str:
         """Record a model load/unload alongside the request traffic that caused it.
 
-        ``running=True`` opens the row (a load in progress) and the caller closes
-        it with the usual :meth:`finish` / :meth:`fail`; an unload is terminal on
-        arrival. Rows are shared, so every subject sees them, and share the same
-        retention budget as requests.
+        ``running=True`` opens the row for the caller to close with :meth:`finish` /
+        :meth:`fail`; an unload is terminal on arrival. Rows are shared (visible to
+        every subject) and share the request retention budget.
         """
         now = time.time()
         entry = ApiMonitorEntry(
@@ -177,8 +176,8 @@ class ApiMonitor:
         return entry.id
 
     def relabel(self, entry_id: Optional[str], model: str) -> None:
-        """Rename an open lifecycle row once the load resolves its real id (the
-        caller only has the load path up front, which may be an HF snapshot dir)."""
+        """Rename an open lifecycle row once the load resolves its real id: up front
+        the caller only has the load path, which may be an HF snapshot dir."""
         if not entry_id or not model:
             return
         with self._lock:
@@ -198,8 +197,7 @@ class ApiMonitor:
                 entry.updated_at = time.time()
 
     def discard(self, entry_id: Optional[str]) -> None:
-        """Drop a row that turned out not to be an event (a load that was already
-        satisfied, so nothing was actually loaded)."""
+        """Drop a row that turned out not to be an event (an already-satisfied load)."""
         if not entry_id:
             return
         with self._lock:
@@ -293,9 +291,8 @@ class ApiMonitor:
             self._trim_terminal_locked()
 
     def fail_open(self, entry_id: Optional[str], error: str) -> None:
-        """Fail only a still-open row. Unlike :meth:`fail` this never touches an
-        entry that already finished, so a catch-all in a ``finally`` cannot stamp
-        an error onto a request that in fact succeeded."""
+        """Fail only a still-open row: unlike :meth:`fail`, a catch-all in a
+        ``finally`` cannot stamp an error onto a request that already succeeded."""
         if not entry_id:
             return
         with self._lock:
diff --git a/studio/backend/core/inference/llama_keepwarm.py b/studio/backend/core/inference/llama_keepwarm.py
index f3ec5f573f..05b1271b27 100644
--- a/studio/backend/core/inference/llama_keepwarm.py
+++ b/studio/backend/core/inference/llama_keepwarm.py
@@ -346,8 +346,8 @@ def _loaded_identity(backend):
 
 
 def _note_idle_unload_event(freed) -> None:
-    """Record an idle auto-unload in the API monitor, using the advertised repo id
-    from the stash so the row never shows the on-disk load path. Best-effort."""
+    """Monitor row for an idle auto-unload. Best-effort; uses the stash's
+    advertised repo id so the row never shows the on-disk load path."""
     try:
         from core.inference.api_monitor import api_monitor
         from core.inference.model_ids import public_model_id
diff --git a/studio/backend/core/inference/local_model_resolver.py b/studio/backend/core/inference/local_model_resolver.py
index c4ac085ebe..5d2a9e9c87 100644
--- a/studio/backend/core/inference/local_model_resolver.py
+++ b/studio/backend/core/inference/local_model_resolver.py
@@ -36,9 +36,8 @@ _lock = threading.Lock()
 _scan: tuple[float, dict[str, _LocalGgufEntry]] = (0.0, {})
 # Not _lock: that is held for the whole scan, so the request path would wait on it.
 _warm_lock = threading.Lock()
-# Repos that finished downloading but are not in the published index yet. The
-# retained index covers what was already known; nothing covers the one that just
-# landed until the next scan, and the request path must not call it absent.
+# Repos that finished downloading but are not in the published index yet: nothing
+# else covers them until the next scan, and the request path must not call them absent.
 _just_downloaded: set[str] = set()
 _warming = False
 _last_scan_s = 0.0
@@ -115,10 +114,9 @@ def _local_gguf_entry(loader_id: str, info) -> Optional[_LocalGgufEntry]:
         quants = tuple(v.quant for v in variants if getattr(v, "quant", None))
         if not quants:
             return None
-        # That call orders by descending size, so the head is the biggest quant,
-        # often F16. A bare id means whichever quant a plain load would take, so put
-        # that first: everything downstream reads [0], and answering with the
-        # largest can evict a working model and then OOM starting it.
+        # That call orders by descending size, so the head is the biggest quant (often
+        # F16). Downstream reads [0], and a bare id must mean whichever quant a plain
+        # load would take: answering with the largest can evict a model and then OOM.
         from core.inference.openai_auto_download import preferred_quant
 
         best = preferred_quant(quants)
@@ -132,9 +130,8 @@ def _local_gguf_entry(loader_id: str, info) -> Optional[_LocalGgufEntry]:
 def local_gguf_quants(info) -> Optional[tuple[str, ...]]:
     """On-disk quant labels for *info*, or None when it is not a servable local
     GGUF. Read from the files, not ``info.model_format``: the HF-cache scanner
-    leaves model_format unset for GGUF snapshots, so a model_format filter would
-    drop every cached GGUF. Lets /v1/models advertise exactly what /v1 can serve,
-    and which quant to name, from a single scan."""
+    leaves that unset for GGUF snapshots, so filtering on it drops every cached
+    GGUF. One scan tells /v1/models what it can serve and which quant to name."""
     from pathlib import Path
 
     path = getattr(info, "path", None)
@@ -330,14 +327,14 @@ def recently_downloaded(repo_id: str) -> bool:
 
 
 def invalidate_index() -> None:
-    """Mark the cached scan stale so the next resolve sees a just-finished
-    download, rather than waiting out the TTL.
+    """Mark the cached scan stale so the next resolve sees a just-finished download
+    instead of waiting out the TTL.
 
-    Keeps the entries. Callers on the request path read this cache without
-    scanning, so emptying it would leave them with no evidence about any local
-    model until the rebuild lands, and a bare request for one of them would be
-    answered by whatever is resident. Only a completed download invalidates, and
-    that only ever adds models, so the retained entries stay true.
+    Keeps the entries: the request path reads this cache without scanning, so
+    emptying it would leave it with no evidence about any local model until the
+    rebuild lands, and a bare request for one would be answered by whatever is
+    resident. Only a completed download invalidates, and that only adds, so the
+    retained entries stay true.
     """
     global _scan
     with _lock:
@@ -366,9 +363,9 @@ def _index() -> dict[str, _LocalGgufEntry]:
 def index_is_built() -> bool:
     """Whether a scan has ever completed, freshness aside.
 
-    Lock-free on purpose: ``_lock`` is held for the whole scan, so taking it here
-    would park the request path on the very scan it is trying to stay off. Reading
-    ``_scan[0]`` is safe because ``_scan`` is only ever rebound, never mutated.
+    Lock-free on purpose: ``_lock`` is held for the whole scan, so taking it would
+    park the request path on the scan it is trying to stay off. Safe because
+    ``_scan`` is only ever rebound, never mutated.
     """
     return bool(_scan[0])
 
@@ -376,13 +373,10 @@ def index_is_built() -> bool:
 def warm_index_soon() -> None:
     """(Re)build the index off the request path when it is missing or past its TTL.
 
-    Callers that cannot afford the scan use this plus ``allow_scan=False``, so this
-    is the only thing that ever refreshes the index for them. It has to cover a
-    stale index and not just an absent one: a model downloaded through the Hub UI
-    or dropped into a scan folder has no invalidation hook, and would otherwise stay
-    invisible to those callers for the life of the process.
-
-    Never touches ``_lock``, which the scan holds throughout, and never blocks.
+    The only refresh for callers using ``allow_scan=False``. Covers a stale index,
+    not just an absent one: a model downloaded through the Hub UI or dropped into a
+    scan folder has no invalidation hook and would otherwise stay invisible to them
+    for the life of the process. Never blocks, and never touches ``_lock``.
     """
     global _warming
     if time.monotonic() - _scan[0] < max(_CACHE_TTL_S, _last_scan_s * _WARM_DUTY):
@@ -419,11 +413,10 @@ def resolve_local_gguf(
     off and resolves only when that quant is on disk, unless it names no quant at
     all (an Ollama-style ":latest"), which means the repo.
 
-    ``allow_scan=False`` answers from the last built index and never rebuilds,
-    for callers on the request path: the scan walks several model dirs and HF
-    caches, takes seconds on a large install, and holds a lock every other
-    caller queues behind. A stale answer is fine there, since what is on disk
-    barely moves and a finished download calls :func:`invalidate_index`.
+    ``allow_scan=False`` answers from the last built index and never rebuilds, for
+    the request path: the scan walks several model dirs and HF caches, takes seconds
+    on a large install, and holds a lock everyone queues behind. Stale is fine there,
+    since disk barely moves and a finished download calls :func:`invalidate_index`.
     """
     if not isinstance(requested, str) or not requested.strip():
         return None
@@ -466,10 +459,9 @@ def describe_local_miss(requested: str) -> tuple[str, tuple[str, ...]]:
     """Why :func:`resolve_local_gguf` missed, so an error can say "wrong quant"
     instead of "no such model".
 
-    ``(MISS_VARIANT_NOT_FOUND, <local quants>)`` when the repo is downloaded but
-    the requested ``:VARIANT`` is not, else ``(MISS_MODEL_NOT_FOUND, ())``. Splits
-    the name like the resolver so the two agree. Fail-safe: a scan failure reports
-    the generic miss rather than raising into the handler.
+    ``(MISS_VARIANT_NOT_FOUND, <local quants>)`` when the repo is downloaded but the
+    requested ``:VARIANT`` is not, else ``(MISS_MODEL_NOT_FOUND, ())``. Fail-safe: a
+    scan failure reports the generic miss rather than raising into the handler.
     """
     if not isinstance(requested, str) or not requested.strip():
         return MISS_MODEL_NOT_FOUND, ()
diff --git a/studio/backend/core/inference/model_ids.py b/studio/backend/core/inference/model_ids.py
index a6270b955e..3886307ae2 100644
--- a/studio/backend/core/inference/model_ids.py
+++ b/studio/backend/core/inference/model_ids.py
@@ -42,9 +42,8 @@ def _looks_like_path(identifier: str) -> bool:
 def hf_cache_repo_id(path: Optional[str]) -> Optional[str]:
     """``.../models--org--name/snapshots/<sha>`` -> ``org/name``, else None.
 
-    A model loaded straight out of the HF cache has a snapshot directory as its
-    identifier, whose basename is a commit hash. Recover the repo id so callers
-    show ``unsloth/gemma-4-31B-it-GGUF`` rather than ``c1ac76e99d55...``.
+    A model loaded from the HF cache is identified by its snapshot dir, whose
+    basename is a commit hash; recover the repo id so callers don't show that.
     """
     if not path:
         return None
diff --git a/studio/backend/core/inference/openai_auto_download.py b/studio/backend/core/inference/openai_auto_download.py
index fee87a42f2..cad5e40d14 100644
--- a/studio/backend/core/inference/openai_auto_download.py
+++ b/studio/backend/core/inference/openai_auto_download.py
@@ -5,22 +5,21 @@
 
 Auto-switch only loads models already on disk. With
 ``openai_api_auto_download_model`` on, a miss that looks like a real Hub repo is
-downloaded in the background instead of erroring, and the request is told to
-retry rather than being held open: a quant is routinely tens of GB, far longer
-than any client (or the Cloudflare edge on ``--secure``) will wait, and the
-inference lifecycle gate must not be held meanwhile. The resident model keeps
-serving throughout, and the retry that lands after the download is served by the
-new model through the ordinary auto-switch path.
+fetched in the background and the request is told to retry rather than held
+open: a quant is routinely tens of GB, far longer than any client (or the
+Cloudflare edge on ``--secure``) will wait, and the inference lifecycle gate must
+not be held meanwhile. The resident model keeps serving, and the retry that lands
+after the download goes through the ordinary auto-switch path.
 
 Admission is deliberately narrow, since a request only needs an API key:
-- ``namespace/name`` only, and only when the Hub confirms it is a GGUF repo.
-  ``gpt-4`` and ``anthropic/claude-3.5-sonnet`` alike fall through to the
-  resident model as before: a namespace is not evidence of intent, since LiteLLM
-  and OpenRouter address every provider that way.
-- GGUF repos only, decided from the remote file list, not the repo name. GGUF
-  runs under llama.cpp, which never imports repo Python.
-- Anything declaring ``auto_map`` is refused, so ``trust_remote_code`` can only
-  ever be granted deliberately in the UI, never by an API call.
+- ``namespace/name`` only, and only when the Hub confirms GGUF weights. A
+  namespace is not evidence of intent (LiteLLM and OpenRouter address every
+  provider that way), so ``gpt-4`` and ``anthropic/claude-3.5-sonnet`` alike
+  fall through to the resident model as before.
+- GGUF only, decided from the remote file list, not the repo name: GGUF runs
+  under llama.cpp, which never imports repo Python.
+- ``auto_map`` is refused, so ``trust_remote_code`` is only ever granted
+  deliberately in the UI, never by an API call.
 - One download at a time, so a key holder cannot fan out fetches.
 """
 
@@ -39,30 +38,29 @@ logger = get_logger(__name__)
 
 # Keep the Hub probe short so a slow Hub can't stall the request path.
 _MODEL_INFO_TIMEOUT_S = 8.0
-# auth_check and hf_hub_download take no timeout of their own, and both run while the
-# provisional slot is held, so an unresponsive Hub would pin the single flight and stall
-# the request long past the metadata budget. The code probe fetches up to three small
-# configs, so it gets more room than the single auth call.
+# auth_check and hf_hub_download take no timeout of their own and run while the
+# provisional slot is held, so an unresponsive Hub would pin the single flight. The
+# code probe fetches up to three configs, so it gets more room than the auth call.
 _CODE_PROBE_TIMEOUT_S = 20.0
 # Headroom left free after the download, so filling the disk can't wedge the box.
 _DISK_RESERVE_BYTES = 5 * 1024**3
 _WATCH_POLL_S = 2.0
 # A stalled watcher must not pin the single-flight slot forever.
 _MAX_WATCH_S = 24 * 60 * 60
-# Past the watch window the row is already resolved, so poll only to see whether
-# the worker is still alive and still owns the slot.
+# Past the watch window the row is resolved, so poll only to see whether the
+# worker is still alive and still owns the slot.
 _TIMED_OUT_POLL_S = 60.0
 _RETRY_AFTER_S = 30
 # Long enough for a client honouring Retry-After to come back and be told, short
-# enough that a client that never returns cannot hold the slot.
+# enough that one that never returns cannot hold the slot.
 _FAILED_HOLD_S = 3 * _RETRY_AFTER_S
 _MAX_LISTED_VARIANTS = 8
 
 
 @dataclass(frozen = True)
 class AutoDownloadRefusal:
-    """Why this request cannot be served yet. The route turns it into an
-    HTTPException with the surface's own error envelope."""
+    """Why this request cannot be served yet; the route raises it in the
+    surface's own error envelope."""
 
     status: int
     code: str
@@ -78,9 +76,8 @@ class _Active:
     expected_bytes: int = 0
     monitor_id: Optional[str] = None
     started_at: float = 0.0
-    # Set when the worker failed. The slot is kept until a retry surfaces it, since
-    # the advertised retry interval is far longer than the watcher's poll and the
-    # client would otherwise just restart the same failing download.
+    # Set when the worker failed. Held until a retry surfaces it: Retry-After is far
+    # longer than the watcher poll, so the client would restart the same failing download.
     error: Optional[str] = None
     failed_at: float = 0.0
 
@@ -103,9 +100,8 @@ def split_model_ref(requested: str) -> tuple[str, Optional[str]]:
     """``org/repo:QUANT`` -> ``("org/repo", "QUANT")``; no suffix -> variant None.
 
     Splits on the last colon. A slash-bearing suffix is only a variant when a real
-    Hub repo precedes it: an unrecognized GGUF below a subdirectory keys on its path
-    ("build/llama-13b", which is_valid_gguf_variant allows and the catalog advertises),
-    while "C:/models/x.gguf" leaves a drive letter that is no repo id at all.
+    Hub repo precedes it: "build/llama-13b" is a subdirectory GGUF key the catalog
+    advertises, while "C:/models/x.gguf" leaves a drive letter that is no repo id.
     """
     text = (requested or "").strip()
     base, sep, suffix = text.rpartition(":")
@@ -122,9 +118,9 @@ def split_model_ref(requested: str) -> tuple[str, Optional[str]]:
 def is_downloadable_ref(requested: str) -> bool:
     """Whether *requested* is shaped like a Hub repo we may fetch.
 
-    Requires an explicit namespace. That keeps ``gpt-4`` and other foreign ids
-    falling through untouched, and avoids the bare-name ``unsloth/`` prefixing in
-    ModelConfig.from_identifier turning an unrelated label into a real repo.
+    Requires an explicit namespace: keeps ``gpt-4`` and other foreign ids falling
+    through, and stops ModelConfig.from_identifier's bare-name ``unsloth/``
+    prefixing from turning an unrelated label into a real repo.
     """
     from hub.utils.paths import is_valid_repo_id
 
@@ -140,9 +136,9 @@ def is_downloadable_ref(requested: str) -> bool:
 def looks_like_quant(variant: Optional[str]) -> bool:
     """Whether a ``:suffix`` names a GGUF quant rather than a foreign tag.
 
-    ``vendor/model`` is how LiteLLM and OpenRouter address every provider, and
-    ``name:latest`` is how Ollama tags one, so neither a namespace nor a colon
-    proves a request was meant for this server. A real quant label does.
+    Neither a namespace nor a colon proves a request was meant for this server
+    (``vendor/model`` is LiteLLM/OpenRouter, ``name:latest`` is Ollama). A real
+    quant label does.
     """
     import re
 
@@ -156,20 +152,16 @@ def looks_like_quant(variant: Optional[str]) -> bool:
 
 
 def _hub_token(hf_token: Optional[str]):
-    """The caller's token, or an explicit False.
-
-    None makes huggingface_hub fall back to a cached login, which here would be
-    the server owner's. False is what actually means anonymous.
-    """
+    """The caller's token, or an explicit False. None makes huggingface_hub fall
+    back to a cached login (here the server owner's); only False is anonymous."""
     return hf_token or False
 
 
 def _servable_key(repo_id: str, hf_token: Optional[str]) -> str:
     """Cache key, per credential.
 
-    The Hub answers 404 for a private repo the caller cannot see, so a verdict
-    reached without a token says nothing about a caller who has one. Keyed on a
-    digest so the token itself is never held here.
+    The Hub 404s a private repo the caller cannot see, so a tokenless verdict says
+    nothing about a caller who has one. Digested, so no token is held here.
     """
     import hashlib
 
@@ -212,8 +204,7 @@ async def _bounded_probe(fn, *args, timeout: float, default):
     """Run a blocking Hub probe off the loop, bounding only the wait.
 
     The thread is left to finish (a blocking socket read cannot be cancelled); the
-    caller stops waiting and takes *default*, which each call site chooses so that a
-    timeout errs the safe way.
+    caller takes *default*, chosen per call site so a timeout errs the safe way.
     """
     try:
         return await asyncio.wait_for(asyncio.to_thread(fn, *args), timeout)
@@ -239,10 +230,9 @@ def _gguf_variants(siblings) -> dict[str, int]:
     """Quant label -> bytes the download will actually fetch.
 
     Mirrors list_gguf_variants for the selectable labels: companions (mmproj/MTP)
-    and big-endian builds are not quants of their own, and sharded quants sum
-    across their shards. The byte total comes from the download plan, which folds
-    the companions back into every quant, so the disk reserve is measured against
-    what the worker fetches rather than the main files alone.
+    and big-endian builds are not quants, and sharded quants sum across shards.
+    Bytes come from the download plan, which folds companions back into every
+    quant, so the disk reserve is measured against what the worker fetches.
     """
     from hub.utils.gguf import extract_quant_label as canonical_quant_label
     from hub.utils.gguf_plan import build_gguf_variant_plans
@@ -262,10 +252,9 @@ def _gguf_variants(siblings) -> dict[str, int]:
             continue
         quant = _extract_quant_label(name)
         if not looks_like_quant(quant):
-            # With no recognized quant token the two extractors part ways: this one
-            # takes the last hyphenated segment ("7b" of llama-7b) while the plan and
-            # the worker key the whole stem. Advertising ours dispatches a variant the
-            # worker cannot resolve, so take theirs for the unrecognized case only.
+            # With no recognized quant token the extractors part ways: this one takes
+            # the last hyphenated segment ("7b" of llama-7b) while the plan and worker
+            # key the whole stem, so advertising ours dispatches an unresolvable variant.
             quant = canonical_quant_label(name) or quant
         if _is_mmproj(name) or _is_mtp_drafter(name) or _is_big_endian_gguf_path(name, quant):
             continue
@@ -343,11 +332,9 @@ async def _progress_percent(
 def _release(active: Optional[_Active]) -> None:
     """Free the single-flight slot, but only while *active* still owns it.
 
-    Keying the release on ``repo_id`` alone let a stale operation clear a newer
-    one for the same repo: variant A errors, an adopting request frees the slot,
-    a retry starts variant B, and A's watcher then matches on the repo and clears
-    B on its way out -- admitting a second repository download alongside B.
-    Identity ties every release to the operation that actually took the slot.
+    Keying on ``repo_id`` alone let a stale operation clear a newer one: variant A
+    errors, an adopting request frees the slot, a retry starts B, then A's watcher
+    matches the repo and clears B, admitting a second download alongside it.
     """
     global _active
     if active is None:
@@ -371,10 +358,9 @@ async def _watch(active: _Active, hf_token: Optional[str]) -> None:
             state, error = await _job_state(active.repo_id, active.variant)
             if state in ("running", "cancelling", "unknown"):
                 if timed_out:
-                    # A worker still running still owns the slot: releasing it on the
-                    # clock alone would admit a second multi-GB download alongside it.
-                    # "unknown" cannot confirm it is alive, so stop holding it then,
-                    # or a broken probe would wedge auto-download for good.
+                    # A running worker still owns the slot: releasing on the clock alone
+                    # would admit a second multi-GB download beside it. "unknown" cannot
+                    # confirm it is alive, so release then, or a broken probe wedges us.
                     if state == "unknown":
                         return
                     continue
@@ -396,16 +382,16 @@ async def _watch(active: _Active, hf_token: Optional[str]) -> None:
                 return
             if state == "complete":
                 # No invalidate here: finalize_worker_exit already dropped the cache and
-                # started the warm, and a second one would mark that fresh scan stale and
-                # push a synchronous rescan onto the client's retry.
+                # warmed it; a second would mark that fresh scan stale and push a
+                # synchronous rescan onto the client's retry.
                 api_monitor.finish(active.monitor_id, status = "completed")
             elif state == "idle":
                 # The job vanished without a terminal state (worker killed).
                 api_monitor.fail_open(active.monitor_id, "Download did not complete")
             else:
                 api_monitor.fail_open(active.monitor_id, error or f"Download {state}")
-                # Keep the slot so the next retry is told it failed rather than
-                # silently starting the same download again.
+                # Keep the slot so the next retry is told it failed instead of
+                # silently restarting the same download.
                 active.error = error or f"Download {state}"
                 active.failed_at = time.monotonic()
                 return
@@ -434,9 +420,8 @@ async def _is_downloadable_model(repo_id: str, hf_token: Optional[str]) -> bool:
     """Whether the Hub has this repo with GGUF weights we could fetch.
 
     Only asked while another download holds the slot, to tell a second download
-    apart from an ordinary foreign label. Any failure answers False: falling
-    through to the resident model is what such a label does anyway, and refusing
-    it would strand normal traffic for the length of the download.
+    apart from an ordinary foreign label. Any failure answers False: refusing
+    would strand normal traffic for the length of the download.
     """
     if _is_not_servable(repo_id, hf_token):
         return False
@@ -450,9 +435,8 @@ async def _is_downloadable_model(repo_id: str, hf_token: Optional[str]) -> bool:
     except Exception:
         return False
     # The same filter admission uses, not a bare .gguf test: mmproj, MTP drafters and
-    # big-endian builds are companions rather than quants, so a repo holding only those
-    # is not downloadable here either. Answering otherwise would hold an ordinary
-    # foreign label at model_download_busy for the length of an unrelated download.
+    # big-endian builds are companions, not quants. Answering otherwise would hold an
+    # ordinary foreign label at model_download_busy for an unrelated download.
     servable = bool(_gguf_variants(getattr(info, "siblings", None)))
     if not servable:
         _mark_not_servable(repo_id, hf_token)
@@ -470,9 +454,9 @@ async def maybe_auto_download(
     Returns None when the request should carry on unchanged, or a refusal the
     caller must raise. Only called after the local resolver has already missed.
 
-    ``require_vision`` refuses a target with no mmproj companion rather than
-    spending gigabytes on weights that cannot answer the request that asked for
-    them; the local capability guard only ever sees an already-downloaded model.
+    ``require_vision`` refuses a target with no mmproj companion rather than spend
+    gigabytes on weights that cannot answer the request; the local capability guard
+    only ever sees an already-downloaded model.
     """
     global _active
 
@@ -502,9 +486,8 @@ async def maybe_auto_download(
 
     if busy is not None:
         # Refusing before the probe blocks ordinary drop-in traffic: a namespaced label
-        # that is not a downloadable GGUF repo (LiteLLM/OpenRouter style) would be told
-        # to wait out a multi-hour download instead of falling through to the resident
-        # model. Only a label that could itself be downloaded is a second download.
+        # that is no downloadable GGUF repo (LiteLLM/OpenRouter style) would be told to
+        # wait out a multi-hour download. Only a downloadable label is a 2nd download.
         if not await _is_downloadable_model(repo_id, hf_token):
             return None
         return AutoDownloadRefusal(
@@ -636,8 +619,7 @@ async def _admit_and_start(
 
     # _hub_token, not the raw token: None lets huggingface_hub fall back to a cached
     # server login, so a caller-named repo would be probed with this server's identity.
-    # Same rule as the metadata probe and the worker.
-    # None on timeout, which refuses: an unchecked repo is not a cleared one.
+    # Defaults to None on timeout, which refuses: unchecked is not cleared.
     has_auto_map = await _bounded_probe(
         _config_has_auto_map,
         repo_id,
@@ -715,8 +697,8 @@ def preferred_quant(labels) -> Optional[str]:
     """The quant a plain load would pick from *labels*, or None.
 
     The one ranking for "which quant did they mean": local resolution, remote
-    admission and what /v1/models advertises all have to agree, or a bare id
-    means a different quant depending on which of them answered it.
+    admission and /v1/models must agree, or a bare id means a different quant
+    depending on which of them answered it.
     """
     from utils.models.model_config import _pick_best_gguf
 
@@ -732,15 +714,14 @@ def _match_variant(wanted: Optional[str], variants: dict[str, int]) -> Optional[
     """Resolve the requested quant against what the repo actually has.
 
     An explicit quant matches case-insensitively and must exist: never quietly
-    substitute another, unlike the loader's low-disk fallback. A bare repo id, or
-    an Ollama-style tag that names no quant at all (":latest", ":8b"), uses the
-    same preference order as a manual load, matching what the local resolver does
-    with the same tag.
+    substitute another, unlike the loader's low-disk fallback. A bare repo id, or a
+    tag that names no quant (":latest", ":8b"), uses the same preference order as a
+    manual load, matching what the local resolver does with the same tag.
     """
     if wanted:
-        # Exact first, whatever shape it is: a repo of generically named GGUFs has
-        # real variants like "llama-13b" that are valid worker keys but do not look
-        # like quants, and defaulting past one would fetch a model nobody asked for.
+        # Exact first, whatever shape: a repo of generically named GGUFs has real
+        # variants like "llama-13b" that are valid worker keys but not quant-shaped,
+        # and defaulting past one would fetch a model nobody asked for.
         lowered = {name.lower(): name for name in variants}
         exact = lowered.get(wanted.strip().lower())
         if exact is not None or looks_like_quant(wanted):
diff --git a/studio/backend/hub/services/download_lifecycle.py b/studio/backend/hub/services/download_lifecycle.py
index b01e25500e..f431f4497e 100644
--- a/studio/backend/hub/services/download_lifecycle.py
+++ b/studio/backend/hub/services/download_lifecycle.py
@@ -241,11 +241,10 @@ def finalize_worker_exit(
     state = classify_exit(rc, cancel_requested = cancel_requested)
     if state == "complete":
         registry.set_job(key, "complete")
-        # Where /v1 learns a new model exists: its resolver answers the request path
-        # from a cached scan with no watcher, and would otherwise report the model
-        # absent and let the request be served by whatever is resident. Models only,
-        # since datasets share this path and noting one as a local model would refuse
-        # a bare request naming that id instead of letting a foreign id fall through.
+        # Where /v1 learns a new model exists: its resolver answers from a cached scan
+        # with no watcher, so it would report the model absent and serve whatever is
+        # resident. Models only: noting a dataset id as a local model would refuse a
+        # bare request naming it instead of letting a foreign id fall through.
         if repo_type == "model":
             try:
                 from core.inference.local_model_resolver import (
@@ -256,8 +255,8 @@ def finalize_worker_exit(
 
                 note_downloaded(repo_id)
                 invalidate_index()
-                # Rebuild here rather than on the first request that needs it, so the
-                # new model resolves without a scan on the request path.
+                # Rebuild here, not on the first request, to keep the scan off the
+                # request path.
                 warm_index_soon()
             except Exception:
                 pass
diff --git a/studio/backend/hub/services/models/downloads.py b/studio/backend/hub/services/models/downloads.py
index 26eb15cb9f..5dc4ebef37 100644
--- a/studio/backend/hub/services/models/downloads.py
+++ b/studio/backend/hub/services/models/downloads.py
@@ -114,8 +114,8 @@ async def download_model_response(
 ):
     """Start a background download for a HuggingFace model.
 
-    ``allow_ambient_token=False`` keeps the worker anonymous when the caller
-    supplied no token, for repos named over the API rather than chosen here.
+    ``allow_ambient_token=False`` keeps the worker anonymous when the caller sent
+    no token, for repos named over the API rather than chosen here.
     """
     repo_id = body.repo_id.strip()
     if not _is_valid_repo_id(repo_id):
diff --git a/studio/backend/routes/inference.py b/studio/backend/routes/inference.py
index ec2ca5365a..0b0d3110f1 100644
--- a/studio/backend/routes/inference.py
+++ b/studio/backend/routes/inference.py
@@ -3118,8 +3118,8 @@ def _lifecycle_model_label(model: Optional[str], variant: Optional[str] = None)
 def _close_load_event(
     entry_id: Optional[str], model: Optional[str], variant: Optional[str]
 ) -> None:
-    """Close a monitor load row, relabelled with the id the load actually resolved
-    (the row opened on the request's model_path, which may be an HF snapshot dir)."""
+    """Close a monitor load row, relabelled with the id the load resolved: the row
+    opened on the request's model_path, which may be an HF snapshot dir."""
     api_monitor.relabel(entry_id, _lifecycle_model_label(model, variant))
     api_monitor.finish(entry_id)
 
@@ -3127,8 +3127,8 @@ def _close_load_event(
 def _monitor_active_model() -> Optional[str]:
     """The loaded model as a client-facing id, quant included when known.
 
-    Cleaned like /v1/models: this is rendered in the settings UI and served over
-    the public --secure tunnel, so it must never be the on-disk load path.
+    Cleaned like /v1/models: rendered in the settings UI and served over the public
+    --secure tunnel, so it must never be the on-disk load path.
     """
     llama_backend = get_llama_cpp_backend()
     if getattr(llama_backend, "is_loaded", False):
@@ -3540,8 +3540,8 @@ _DISABLE_OPENAI_AUTO_SWITCH_SCOPE_KEY = "_unsloth_disable_openai_auto_switch"
 # only restore an idle-freed model, never run the resolver (so a downloaded GGUF
 # literally named "default" can't be swapped to). The NUL keeps it off any index.
 _RELOAD_ONLY_MODEL = "\x00reload-only"
-# One cold scan is worth paying to avoid answering a named model with another; a
-# pathological install must not hang the request behind it forever.
+# One cold scan is worth paying to avoid answering a named model with another,
+# bounded so a pathological install cannot hang the request behind it.
 _COLD_INDEX_WAIT_S = 10.0
 
 
@@ -3659,9 +3659,9 @@ def _format_available_models(ids: list[str]) -> str:
 async def _unavailable_model_message(requested_model: str) -> str:
     """Why a named model can't serve this request, and what can.
 
-    Auto-switch only loads already-downloaded GGUFs, so a request naming a real
-    model usually fails because it is not on this machine. Pointing the caller at
-    /inference/load cannot fix that; say what is actually wrong.
+    Auto-switch only loads downloaded GGUFs, so a request naming a real model
+    usually fails because it is not on this machine, which /inference/load cannot
+    fix; say what is actually wrong.
     """
     from core.inference.local_model_resolver import (
         MISS_VARIANT_NOT_FOUND,
@@ -3694,10 +3694,10 @@ async def _no_model_loaded_error(
 ):
     """``(status, detail)`` for the /v1 sites that fail because nothing is loaded.
 
-    Changes only the case the generic text describes wrongly: auto-switch on, a
-    model named, and that name resolves to nothing local, so the switch silently
-    did nothing. That becomes a 404 model_not_found. Toggle off or no model named
-    keeps ``status`` and the :func:`_no_model_loaded_detail` text verbatim.
+    Changes only the case the generic text gets wrong (auto-switch on, a model
+    named, that name resolving to nothing local, so the switch silently did
+    nothing) into a 404 model_not_found. Everything else keeps ``status`` and the
+    :func:`_no_model_loaded_detail` text verbatim.
     """
     from utils.openai_auto_switch_settings import get_openai_auto_switch_enabled
     from core.inference.local_model_resolver import resolve_local_gguf
@@ -3739,10 +3739,9 @@ async def _no_model_loaded_error(
 def _auto_download_hf_token(fastapi_request: Optional[Request]) -> Optional[str]:
     """The token to fetch with: only one the caller sent themselves.
 
-    Never the server's ambient token. The repo here is named by whoever holds an
-    API key, so borrowing the owner's Hub identity would let that key pull the
-    owner's private repos and publish them in /v1/models for every other key.
-    The OpenAI bearer key is never used as an HF token either.
+    Never the server's ambient token, and never the OpenAI bearer key. The repo is
+    named by whoever holds an API key, so borrowing the owner's Hub identity would
+    let that key pull the owner's private repos and publish them in /v1/models.
     """
     from hub.dependencies import HUB_HF_TOKEN_HEADER, HUB_HF_TOKEN_MAX_LENGTH
 
@@ -3763,10 +3762,9 @@ async def _maybe_auto_download_model(
 ) -> None:
     """Opt-in: start fetching a named GGUF this server doesn't have.
 
-    Raises to stop the request when the model is downloading or cannot be
-    fetched. Off by default, and it never fires on a name that isn't shaped like
-    a Hub repo, so an unknown id like "gpt-4" still falls through to the resident
-    model as before.
+    Raises to stop the request while the model is downloading or cannot be fetched.
+    Off by default, and never fires on a name not shaped like a Hub repo, so an
+    unknown id like "gpt-4" still falls through to the resident model.
     """
     from utils.openai_auto_switch_settings import get_openai_auto_download_enabled
     from core.inference.openai_auto_download import is_downloadable_ref, maybe_auto_download
@@ -3812,8 +3810,8 @@ async def _maybe_auto_download_model(
 def _loaded_satisfies(requested: str) -> bool:
     """Whether what is serving right now actually answers to *requested*.
 
-    A bare ``org/model`` is satisfied by any loaded quant of that repo; an
-    explicit ``:QUANT`` must match the loaded one.
+    A bare ``org/model`` is satisfied by any loaded quant of that repo; an explicit
+    ``:QUANT`` must match the loaded one.
     """
     from core.inference.openai_auto_download import looks_like_quant, split_model_ref
 
@@ -3866,8 +3864,7 @@ def _matches_any(requested: str, candidates) -> bool:
     """Whether *requested* names any of *candidates*.
 
     A repo alias is case-insensitive, a filesystem path is not: lowercasing both
-    made /srv/models/foo.gguf and /srv/models/Foo.gguf the same weights, which is
-    the same trap _norm_path exists for one comparison further down.
+    made /srv/models/foo.gguf and /srv/models/Foo.gguf the same weights.
     """
     lowered = requested.strip().lower()
     for candidate in candidates:
@@ -3893,9 +3890,8 @@ def _norm_path(value: str) -> str:
     /srv/models/Foo and /srv/models/foo are different models."""
     import os
 
-    # normcase after, not before: on Windows it folds case *and* rewrites the
-    # separator to a backslash, so normalizing first leaves the descendant checks
-    # below comparing a "/" against a path that no longer has any.
+    # normcase after, not before: on Windows it folds case *and* rewrites "/" to a
+    # backslash, leaving the descendant checks below comparing against a path with none.
     return os.path.normcase(str(value)).replace("\\", "/").rstrip("/")
 
 
@@ -3908,10 +3904,9 @@ def _resident_quant_is(variant: Optional[str]) -> bool:
 def _resolves_to_resident(load_path: Optional[str], *, llama_only: bool = False) -> bool:
     """Whether a resolved on-disk path is what is already loaded.
 
-    ``llama_only`` drops the Transformers backend from the comparison. Only
-    llama.cpp carries a quant identity, so a Transformers model active from a
-    directory that also holds GGUF exports would otherwise match a request for
-    one of those quants and answer it with the safetensors weights.
+    ``llama_only`` drops the Transformers backend: only llama.cpp carries a quant
+    identity, so a Transformers model active from a directory that also holds GGUF
+    exports would otherwise answer a request for one of those quants.
     """
     if not load_path:
         return False
@@ -3933,9 +3928,9 @@ def _resolves_to_resident(load_path: Optional[str], *, llama_only: bool = False)
             return True
         if current.startswith(f"{target}/"):
             # A model directory holding the weights loaded from it. Nested entries
-            # (/models/A alongside /models/A/sub/B) satisfied this too, so a request
-            # for A was answered with B. The innermost indexed model owns the file;
-            # with none indexed there is no nesting to tell apart, so keep matching.
+            # (/models/A alongside /models/A/sub/B) matched too, so a request for A was
+            # answered with B. The innermost indexed model owns the file; with none
+            # indexed there is no nesting to tell apart, so keep matching.
             owner = _innermost_indexed_owner(current)
             if owner is None or owner == target:
                 return True
@@ -3965,13 +3960,11 @@ async def _reject_unservable_model(
     """Refuse rather than answer a named model with a different one.
 
     Only for a reference this server can tell was meant for it: an explicit GGUF
-    quant, or a model that is actually here. A namespace decides nothing either
-    way. ``vendor/model`` is how LiteLLM and OpenRouter name every provider, so
-    ``anthropic/claude-3.5-sonnet`` falls through like ``gpt-4``; a standalone or
-    custom-folder GGUF is advertised without one, so a slashless id that does
-    resolve locally is still a concrete reference. Only runs while something is
-    serving; with nothing loaded the caller's own :func:`_no_model_loaded_error`
-    already says the right thing.
+    quant, or a model that is actually here. A namespace decides nothing either way
+    (``vendor/model`` is how LiteLLM and OpenRouter name every provider, and a
+    standalone GGUF is advertised without one), so a slashless id that resolves
+    locally is still a concrete reference. Only runs while something is serving:
+    with nothing loaded, :func:`_no_model_loaded_error` already says the right thing.
     """
     from core.inference.openai_auto_download import looks_like_quant, split_model_ref
 
@@ -4006,27 +3999,24 @@ async def _reject_unservable_model(
             warm_index_soon()
             resolved = resolve_local_gguf(requested_model, allow_scan = False)
         else:
-            # Before the first scan there is nothing cached to reason from, and falling
-            # through would answer a named model with the resident one. Pay the scan
-            # once, off the loop and bounded, rather than reading "not scanned yet" as
-            # "not here". Later requests take the cached branch above.
+            # Nothing cached to reason from yet, and falling through would answer a
+            # named model with the resident one. Pay the scan once, off the loop and
+            # bounded, rather than read "not scanned yet" as "not here".
             try:
                 resolved = await asyncio.wait_for(
                     asyncio.to_thread(resolve_local_gguf, requested_model),
                     _COLD_INDEX_WAIT_S,
                 )
             except (TimeoutError, asyncio.TimeoutError):
-                # Still scanning, so nothing is known about this name. Falling through
-                # would put the resident model behind it, which is the failure this
-                # whole hook exists to stop, so say "not yet" instead of guessing.
+                # Still scanning, so nothing is known about this name: say "not yet"
+                # rather than guess and put the resident model behind it.
                 warm_index_soon()
                 still_indexing = True
                 resolved = None
-        # A manual load stores the on-disk path the resolver advertises under an alias, so
-        # match on the path too.
-        # Quants of one repo share a directory, so the path alone cannot tell them
-        # apart: without the variant check an explicit :Q8_0 would be answered by a
-        # resident Q4_K_M, which _loaded_satisfies has already refused by name.
+        # A manual load stores the on-disk path the resolver advertises under an alias,
+        # so match on the path too. Quants of one repo share a directory, so the path
+        # alone cannot tell them apart: without the variant check an explicit :Q8_0
+        # would be answered by a resident Q4_K_M.
         if (
             resolved is not None
             and _resolves_to_resident(resolved[0], llama_only = quantified)
@@ -4052,8 +4042,8 @@ async def _reject_unservable_model(
         )
         switchable = downloaded and get_openai_auto_switch_enabled()
     except HTTPException:
-        # A refusal decided above is the answer, not a failure to decide. Without this
-        # the handler below would log it and fall through to the resident model.
+        # A refusal decided above is the answer, not a failure to decide: without this
+        # the handler below logs it and falls through to the resident model.
         raise
     except Exception as exc:
         # Can't verify: an explicit quant still proves intent, so refuse; let anything else by.
@@ -4189,9 +4179,9 @@ async def _maybe_auto_switch_model(
         # repo, so it never reloads a different local quant that already serves it.
         from core.inference.openai_auto_download import looks_like_quant, split_model_ref
 
-        # A tag that names no quant (":latest", ":8b") means the repo, exactly as
-        # _loaded_satisfies and the resolver read it. Treating it as a quant tears
-        # down a serving Q8 to load the preferred Q4 for a request either satisfies.
+        # A tag that names no quant (":latest", ":8b") means the repo, as
+        # _loaded_satisfies and the resolver read it. Treating it as a quant tears down
+        # a serving Q8 to load the preferred Q4 for a request either satisfies.
         _, _requested_variant = split_model_ref(requested_model)
         bare = not looks_like_quant(_requested_variant)
 
@@ -11113,8 +11103,8 @@ def _quant_reference_resolves(model_id: Optional[str], quant: str) -> bool:
     """Whether ``<model_id>:<quant>`` still resolves once this model is not resident.
 
     A standalone .gguf takes its quant from the filename, but the resolver stores
-    such files with no quants at all, so advertising one hands out a pin that dies
-    the moment another model loads.
+    such files with no quants, so advertising one hands out a pin that dies the
+    moment another model loads.
     """
     from core.inference.local_model_resolver import (
         index_is_built,
@@ -11125,7 +11115,7 @@ def _quant_reference_resolves(model_id: Optional[str], quant: str) -> bool:
 
     if not model_id:
         return False
-    # Cold index proves nothing, and publishing on no proof is what hands out the
+    # A cold index proves nothing, and publishing on no proof is what hands out the
     # dead pin; warm so the next response carries the quant.
     warm_index_soon()
     return resolve_local_gguf(f"{model_id}:{quant}", allow_scan = False) is not None
@@ -11135,8 +11125,8 @@ def _advertised_local_path(model: str) -> Optional[str]:
     """On-disk path of *model* if the last /v1/models scan listed it, else None.
 
     Cache-only, never scans. The catalog scans on its own schedule, so it can have
-    advertised a local model the resolver index has not picked up yet; having
-    advertised it is evidence the name means something other than the resident one.
+    advertised a local model the resolver index has not picked up yet, which is
+    evidence the name means something other than the resident one.
     """
     if _ADVERTISED_CACHE["at"] != _CATALOG_CACHE["at"]:
         paths = {}
@@ -11231,17 +11221,16 @@ async def _openai_catalog_objects() -> list[dict]:
             "object": "model",
             "created": _created,
             "owned_by": _OWNED_BY,
-            # A manual load keys the resident entry by path basename while the catalog uses
-            # the alias, so match on the path or the alias reads as not loaded. llama-only:
-            # these entries are advertised as GGUF with a GGUF quant, so a Transformers
-            # model live from a directory that also holds GGUF exports must not mark one
-            # loaded, or the examples pin a quant nothing can serve with switching off.
+            # A manual load keys the resident entry by path basename while the catalog
+            # uses the alias, so match on the path or the alias reads as not loaded.
+            # llama-only: a Transformers model live from a directory that also holds
+            # GGUF exports must not mark one of these GGUF entries loaded, or the
+            # examples pin a quant nothing can serve with switching off.
             "loaded": _resolves_to_resident(getattr(info, "path", None), llama_only = True),
         }
         # The id stays bare for OpenAI compat; a client appends ":<quant>" to pin one.
-        # For the resident model that has to be the quant actually loaded, not the
-        # preferred one on disk, or the listing advertises alias:Q4 as loaded while
-        # Q8 is serving and pinning it 404s.
+        # For the resident model that must be the quant actually loaded, not the
+        # preferred one on disk, or the listing advertises alias:Q4 while Q8 serves.
         resident_quant = getattr(get_llama_cpp_backend(), "hf_variant", None)
         if obj["loaded"] and resident_quant:
             obj["quant"] = resident_quant
diff --git a/studio/backend/tests/conftest.py b/studio/backend/tests/conftest.py
index 37c933ea41..89f612393c 100644
--- a/studio/backend/tests/conftest.py
+++ b/studio/backend/tests/conftest.py
@@ -62,21 +62,19 @@ def pytest_addoption(parser):
 def _no_background_model_scan(monkeypatch):
     """Keep the /v1 admission hook from scanning the real HF cache during tests.
 
-    The hook warms the local-model index on a background thread. That is right in a
-    server and wrong here: it walks the developer's actual caches, which on a large
-    install takes seconds, and the resulting I/O starves the loop under the
-    timing-sensitive streaming tests. Tests that exercise the warm patch it back.
+    The hook warms the local-model index on a background thread: right in a server,
+    wrong here, since it walks the developer's actual caches and the I/O starves the
+    loop under timing-sensitive streaming tests. Warm tests patch it back.
     """
     import time
 
     from core.inference import local_model_resolver
 
     monkeypatch.setattr(local_model_resolver, "warm_index_soon", lambda: None)
-    # Start from a built, empty index. Stubbing only the background warm still left the
-    # cold path walking those caches synchronously inside the admission wait, so on a
-    # large install the assertion became a 503 "still indexing". Tests that want the
-    # cold path set _scan back themselves (and stub the scan). _build_index is left
-    # alone so the tests that call it directly still exercise the real walk.
+    # Start from a built, empty index: stubbing only the warm left the cold path
+    # walking those caches inside the admission wait, so on a large install the
+    # assertion became a 503 "still indexing". Cold-path tests reset _scan themselves;
+    # _build_index is untouched so tests calling it directly still walk for real.
     monkeypatch.setattr(local_model_resolver, "_scan", (time.monotonic(), {}))
 
 
diff --git a/studio/backend/tests/test_openai_auto_download.py b/studio/backend/tests/test_openai_auto_download.py
index c12a254fc3..b1dde175da 100644
--- a/studio/backend/tests/test_openai_auto_download.py
+++ b/studio/backend/tests/test_openai_auto_download.py
@@ -4,9 +4,8 @@
 """Opt-in auto-download of a GGUF a /v1 request names but this server lacks.
 
 No network: huggingface_hub, the consent probe and the Hub download service are
-all mocked. The invariant these guard is that with the setting off nothing here
-runs at all, and with it on a name that isn't shaped like a repo still falls
-through to the resident model.
+all mocked. The invariant: with the setting off nothing here runs at all, and
+with it on a name not shaped like a repo still falls through to the resident model.
 """
 
 import asyncio
@@ -87,10 +86,9 @@ def _gated_error():
 def _hub_error(error_type, status_code: int, message: str):
     """Build a Hub exception across huggingface_hub majors.
 
-    huggingface_hub 1.x made ``response`` a required keyword-only argument, and
-    the project floor is 0.34, so construct positionally and fall back. The
-    positional form carries no response, and hf_error_status reads the status off
-    it for the types that do not encode it in their name, so attach one either way.
+    huggingface_hub 1.x made ``response`` a required keyword-only argument and the
+    project floor is 0.34, so construct positionally and fall back. The positional
+    form carries no response, which hf_error_status reads, so attach one either way.
     """
     try:
         exc = error_type(message)
@@ -113,9 +111,8 @@ def _hub_error(error_type, status_code: int, message: str):
 
 
 def test_the_hub_error_helper_carries_a_status_on_both_majors():
-    # CI runs huggingface_hub 1.x and this box runs 0.x, and only one of the two
-    # constructor shapes works on each. hf_error_status reads the status off the
-    # response, so a helper that silently produced one without it would make an
+    # CI runs huggingface_hub 1.x and this box 0.x, and each takes only one of the
+    # constructor shapes. A helper that silently dropped the response would make an
     # error-mapping test pass here and fail there.
     from hub.utils.hf_errors import hf_error_status
 
@@ -553,10 +550,9 @@ def test_an_unknown_state_still_reports_the_download_to_a_retry(hub, monkeypatch
 
 
 def test_a_hanging_code_probe_does_not_pin_the_slot(hub, monkeypatch):
-    # hf_hub_download and auth_check take no timeout, and both run while the provisional
-    # slot is held, so an unresponsive Hub stalled the request far past the metadata
-    # budget and reported every other model busy meanwhile. Unchecked is not cleared,
-    # so the bounded probe refuses rather than admitting the repo.
+    # hf_hub_download and auth_check take no timeout and run while the provisional slot
+    # is held, so an unresponsive Hub stalled the request and reported every other model
+    # busy. Unchecked is not cleared, so the bounded probe refuses instead of admitting.
     import threading
 
     entered, release = threading.Event(), threading.Event()
@@ -610,10 +606,9 @@ def test_a_hanging_auth_check_falls_through_to_the_download(hub, monkeypatch):
 
 
 def test_a_companion_only_repo_is_not_held_at_busy(hub):
-    # mmproj and MTP files are companions, not quants, so admission classifies such a
-    # repo as non-servable and lets the label fall through to the resident model. The
-    # busy probe accepted any .gguf, which stranded that ordinary traffic behind an
-    # unrelated multi-hour download.
+    # mmproj and MTP files are companions, not quants, so such a repo is non-servable
+    # and falls through to the resident model. The busy probe accepted any .gguf, which
+    # stranded that ordinary traffic behind an unrelated multi-hour download.
     assert _run("unsloth/x-GGUF:UD-Q4_K_XL").code == "model_downloading"
     gb = 1024**3
     hub["info"] = _Info([_Sibling("mmproj-F16.gguf", gb), _Sibling("mtp-model.gguf", gb)])
@@ -1242,9 +1237,8 @@ def test_the_request_path_never_triggers_a_model_index_rescan(monkeypatch):
 
 
 def test_a_cold_index_is_scanned_rather_than_read_as_nothing_here(monkeypatch):
-    # Before the first scan there is no cached evidence, and treating that as "not
-    # downloaded" answers a named local model with the resident one. The scan is paid
-    # once, off the loop; every later request reads the built index instead.
+    # With no cached evidence yet, reading that as "not downloaded" answers a named
+    # local model with the resident one. Pay the scan once, off the loop.
     from core.inference import local_model_resolver as resolver
 
     entry = resolver._LocalGgufEntry("org/other", "/srv/models/org--other", ("Q4_K_M",))
@@ -1280,9 +1274,8 @@ def test_a_cold_index_is_scanned_rather_than_read_as_nothing_here(monkeypatch):
 
 
 def test_a_cold_scan_that_never_finishes_says_so_instead_of_guessing(monkeypatch):
-    # The scan is bounded so a pathological install cannot hold the request open, but
-    # an unfinished scan knows nothing about the name, and falling through would put
-    # the resident model behind it. Answer "not yet", with a Retry-After.
+    # The scan is bounded, but an unfinished one knows nothing about the name, and
+    # falling through would put the resident model behind it: answer "not yet".
     import threading
 
     from core.inference import local_model_resolver as resolver
@@ -1312,8 +1305,8 @@ def test_a_cold_scan_that_never_finishes_says_so_instead_of_guessing(monkeypatch
 
 def test_a_refusal_is_never_swallowed_by_the_cannot_verify_handler(monkeypatch):
     # The checks run inside a broad `except Exception` that turns a failure to decide
-    # into a fallthrough. An HTTPException raised in there is a decision, and was
-    # being logged as a failure and answered by the resident model instead.
+    # into a fallthrough. An HTTPException there is a decision, but was logged as a
+    # failure and answered by the resident model.
     loaded = _Loaded("unsloth/A-GGUF", "UD-Q4_K_XL")
     monkeypatch.setattr(inference_route, "get_llama_cpp_backend", lambda: loaded)
     monkeypatch.setattr(
@@ -1435,9 +1428,8 @@ def test_an_advertised_alias_for_the_resident_weights_is_still_served(monkeypatc
 
 
 def test_a_rejected_token_says_so_instead_of_asking_for_a_retry(hub):
-    # Hugging Face answers an expired or invalid X-Unsloth-HF-Token with 401. Only
-    # 403 and 404 were handled, so it fell through to "could not reach Hugging Face"
-    # with a 503, telling the caller to retry something that cannot start working.
+    # Hugging Face 401s an expired X-Unsloth-HF-Token. Only 403/404 were handled, so it
+    # fell through to a 503 telling the caller to retry something that cannot work.
     from huggingface_hub.utils import HfHubHTTPError
 
     hub["raise"] = _hub_error(HfHubHTTPError, 401, "unauthorized")
@@ -1448,9 +1440,8 @@ def test_a_rejected_token_says_so_instead_of_asking_for_a_retry(hub):
 
 
 def test_an_image_request_does_not_download_a_text_only_model(hub):
-    # The capability guard only ever sees an already-local target, so without this
-    # an image request would spend gigabytes on weights that cannot answer it and
-    # then 400 on every retry.
+    # The capability guard only ever sees an already-local target, so without this an
+    # image request spends gigabytes on weights that then 400 on every retry.
     gb = 1024**3
     hub["info"] = _Info([_Sibling("model-UD-Q5_K_XL.gguf", 5 * gb)])
     refusal = asyncio.run(
@@ -1489,10 +1480,9 @@ def test_two_models_differing_only_in_case_are_not_the_same_weights(monkeypatch)
 
 
 def test_a_quant_request_is_not_satisfied_by_transformers_weights(monkeypatch):
-    # A Transformers model active from a directory that also holds GGUF exports
-    # resolves to that same directory, and the path match let admission answer an
-    # explicit quant with the safetensors weights. Only llama.cpp has a quant
-    # identity, which is why _loaded_satisfies already refuses this by name.
+    # A Transformers model active from a directory that also holds GGUF exports resolves
+    # to that directory, so the path match let admission answer an explicit quant with
+    # the safetensors weights. Only llama.cpp has a quant identity.
     from core.inference import local_model_resolver as resolver
 
     entry = resolver._LocalGgufEntry("alias", "/srv/models/tuned", ("Q4_K_M",))
@@ -1564,9 +1554,8 @@ def test_a_timed_out_download_stops_holding_the_slot_once_unprobeable(monkeypatc
 
 
 def test_a_sibling_quant_in_the_same_directory_is_not_the_resident_one(monkeypatch):
-    # Quants of one repo share a directory, so the path match alone cannot tell
-    # them apart, and an explicit :Q8_0 was answered by a resident Q4_K_M that
-    # _loaded_satisfies had already refused by name.
+    # Quants of one repo share a directory, so the path match alone cannot tell them
+    # apart, and an explicit :Q8_0 was answered by a resident Q4_K_M.
     from core.inference import local_model_resolver as resolver
 
     entry = resolver._LocalGgufEntry("org/model", "/hf/org--model/snap", ("Q4_K_M", "Q8_0"))
@@ -1587,9 +1576,8 @@ def test_a_sibling_quant_in_the_same_directory_is_not_the_resident_one(monkeypat
 
 
 def test_a_remote_tag_that_names_no_quant_picks_the_preferred_one(hub):
-    # ":latest" and ":8b" name no quant, so remote admission must default-select
-    # like a bare repo id instead of 404ing on a quant that never existed. Matches
-    # what the local resolver now does with the same tag.
+    # ":latest" and ":8b" name no quant, so remote admission must default-select like a
+    # bare repo id (as the local resolver does) instead of 404ing on a non-quant.
     assert _run("unsloth/x-GGUF").code == "model_downloading"
     bare_repo, bare_variant, _ = hub["started"][0]
     for tag in (":latest", ":8b"):
@@ -1608,9 +1596,9 @@ def test_a_remote_tag_that_names_no_quant_picks_the_preferred_one(hub):
 
 
 def test_a_generic_gguf_advertises_the_label_the_worker_resolves(hub):
-    # With no recognized quant token the label extractors part ways: one takes the
-    # last hyphenated segment, the plan and the worker key the whole stem. Dispatching
-    # ours made the worker exit with "No GGUF shards matching variant".
+    # With no recognized quant token the extractors part ways: one takes the last
+    # hyphenated segment, the plan and worker key the whole stem. Dispatching ours
+    # made the worker exit with "No GGUF shards matching variant".
     from hub.utils.gguf import extract_quant_label as canonical
     from hub.utils.gguf_plan import build_gguf_variant_plans
 
@@ -1624,9 +1612,9 @@ def test_a_generic_gguf_advertises_the_label_the_worker_resolves(hub):
 
 
 def test_windows_style_paths_still_match_their_own_directory(monkeypatch):
-    # normcase folds case and rewrites the separator to a backslash on Windows, so
-    # normalizing to "/" before it left the descendant checks comparing a "/" against
-    # a path that had none, and a resident model read as a different one.
+    # normcase rewrites "/" to a backslash on Windows, so normalizing before it left the
+    # descendant checks comparing against a path with none, and a resident model read
+    # as a different one.
     import ntpath
 
     monkeypatch.setattr(inference_route.os.path, "normcase", ntpath.normcase)
@@ -1646,9 +1634,8 @@ def test_windows_style_paths_still_match_their_own_directory(monkeypatch):
 
 
 def test_a_bare_request_for_a_just_downloaded_model_is_refused(monkeypatch):
-    # End of the same chain: the note has to reach admission, or a bare request in
-    # the window between the download landing and the scan is served by the resident
-    # model, which is the whole failure this hook exists to stop.
+    # End of the same chain: the note has to reach admission, or a bare request between
+    # the download landing and the scan is served by the resident model.
     from core.inference import local_model_resolver as resolver
 
     monkeypatch.setattr(resolver, "_scan", (time.monotonic(), {}))
@@ -1668,9 +1655,9 @@ def test_a_bare_request_for_a_just_downloaded_model_is_refused(monkeypatch):
 
 
 def test_a_non_quant_tag_does_not_tear_down_a_serving_quant(monkeypatch):
-    # _already_serving split on ":" rather than on whether the suffix names a quant,
-    # so org/model:latest against a serving Q8_0 counted as a quant mismatch and
-    # swapped in the preferred Q4_K_M, for a request either one satisfies.
+    # _already_serving split on ":" rather than on whether the suffix names a quant, so
+    # org/model:latest against a serving Q8_0 counted as a mismatch and swapped in the
+    # preferred Q4_K_M, for a request either one satisfies.
     from core.inference import local_model_resolver as resolver
 
     entry = resolver._LocalGgufEntry("org/model", "/hf/org--model/snap", ("Q4_K_M", "Q8_0"))
@@ -1698,9 +1685,9 @@ def test_a_non_quant_tag_does_not_tear_down_a_serving_quant(monkeypatch):
 
 
 def test_the_trust_probe_never_falls_back_to_the_server_identity(hub, monkeypatch):
-    # huggingface_hub treats None as "use the cached login", so only an explicit
-    # False is anonymous. The metadata probe and the worker already pass one; this
-    # probe did not, so a caller-named repo was read with the server's identity.
+    # huggingface_hub treats None as "use the cached login", so only an explicit False
+    # is anonymous. This probe passed None, so a caller-named repo was read with the
+    # server's identity.
     seen: list = []
 
     def _probe(model_name, hf_token = None):
@@ -1718,9 +1705,8 @@ def test_the_trust_probe_never_falls_back_to_the_server_identity(hub, monkeypatc
 
 
 def test_a_foreign_label_is_not_told_to_wait_for_someone_elses_download(hub):
-    # The busy refusal fired before the probe, so any namespaced label a drop-in
-    # client sends (LiteLLM/OpenRouter style) was told to wait out a download that
-    # has nothing to do with it, for as long as that download runs.
+    # The busy refusal fired before the probe, so any namespaced label a drop-in client
+    # sends (LiteLLM/OpenRouter style) was told to wait out an unrelated download.
     assert _run("unsloth/first-GGUF").code == "model_downloading"
 
     hub["info"] = _Info([_Sibling("README.md", 1024)])  # real repo, no GGUF
@@ -1733,9 +1719,8 @@ def test_a_foreign_label_is_not_told_to_wait_for_someone_elses_download(hub):
 
 
 def test_a_failed_download_keeps_the_slot_until_someone_is_told(monkeypatch):
-    # The watcher freed the slot the moment it saw the error, but Retry-After is 30s
-    # and the poll is 2s, so the client came back to an empty slot and started the
-    # identical failing download again instead of being told it had failed.
+    # The watcher freed the slot on the error, but Retry-After is 30s and the poll 2s,
+    # so the client came back to an empty slot and restarted the same failing download.
     monkeypatch.setattr(auto_dl, "_MAX_WATCH_S", 60.0)
     monkeypatch.setattr(auto_dl, "_WATCH_POLL_S", 0.001)
 
@@ -1782,9 +1767,8 @@ def test_a_completed_download_does_not_restage_the_scan_it_just_warmed(monkeypat
 
 
 def test_an_exact_generic_variant_beats_the_default_pick(hub):
-    # Canonicalizing generic labels made them real worker keys, but the matcher still
-    # read anything non-quant-shaped as a tag, so repo:llama-13b default-selected and
-    # fetched llama-7b instead of the model that was actually asked for.
+    # Canonicalizing generic labels made them real worker keys, but the matcher read
+    # anything non-quant-shaped as a tag, so repo:llama-13b default-selected llama-7b.
     gb = 1024**3
     hub["info"] = _Info([_Sibling("llama-7b.gguf", 4 * gb), _Sibling("llama-13b.gguf", 8 * gb)])
     assert _run("unsloth/generic-GGUF:llama-13b").code == "model_downloading"
diff --git a/studio/backend/tests/test_openai_auto_switch.py b/studio/backend/tests/test_openai_auto_switch.py
index aa08f9fe1a..065eddfe99 100644
--- a/studio/backend/tests/test_openai_auto_switch.py
+++ b/studio/backend/tests/test_openai_auto_switch.py
@@ -23,9 +23,8 @@ from utils import openai_auto_switch_settings as settings
 def _clean_resolver_index():
     """Drop the scan cache around every test.
 
-    The /v1 admission hook warms the index in the background, so without this a
-    test that exercises the hook can publish its own fixture's scan and, inside the
-    TTL, hand it to the next test that expects a fresh one.
+    The /v1 admission hook warms the index in the background, so a test exercising it
+    can publish its fixture's scan and, inside the TTL, hand it to the next test.
     """
     resolver.invalidate_index()
     yield
@@ -4146,11 +4145,10 @@ def test_env_idle_below_floor_is_clamped(monkeypatch):
 
 
 def test_a_tag_that_names_no_quant_resolves_to_the_repo(monkeypatch):
-    # A downloaded but unloaded GGUF asked for as org/model:latest missed the
-    # resolver, so the switch path could not load it: with auto-download on it
-    # probed the Hub and 404d on a quant that was never a quant, and with it off it
-    # refused without switching. A real quant that is not on disk must still miss,
-    # or a swap would serve the wrong weights under the right name.
+    # A downloaded but unloaded GGUF asked for as org/model:latest missed the resolver,
+    # so the switch could not load it (404ing on a quant that was never a quant with
+    # auto-download on, refusing with it off). A real quant that is not on disk must
+    # still miss, or a swap would serve the wrong weights under the right name.
     from core.inference.local_model_resolver import _LocalGgufEntry
 
     import time
@@ -4173,9 +4171,9 @@ def test_a_tag_that_names_no_quant_resolves_to_the_repo(monkeypatch):
 
 
 def test_any_finished_download_drops_the_resolver_cache(monkeypatch):
-    # Only the API auto-download watcher invalidated, so a GGUF fetched in the Hub
-    # UI stayed absent to the cache-only request path and the request was answered
-    # by the resident model instead. Every worker exits through here.
+    # Only the API auto-download watcher invalidated, so a GGUF fetched in the Hub UI
+    # stayed absent to the cache-only request path and the resident model answered.
+    # Every worker exits through here.
     import logging
 
     from hub.services import download_lifecycle
@@ -4222,9 +4220,9 @@ def test_any_finished_download_drops_the_resolver_cache(monkeypatch):
 
 
 def test_invalidating_keeps_the_entries_it_already_had(monkeypatch):
-    # The request path reads this cache without scanning, so emptying it leaves it
-    # with no evidence about any local model until the rebuild lands. Only a
-    # completed download invalidates, and that only adds, so the entries stay true.
+    # The request path reads this cache without scanning, so emptying it leaves no
+    # evidence until the rebuild lands. Only a completed download invalidates, and
+    # that only adds, so the entries stay true.
     import time
 
     entry = resolver._LocalGgufEntry("org/old", "/srv/models/org--old", ("Q4_K_M",))
@@ -4240,9 +4238,8 @@ def test_invalidating_keeps_the_entries_it_already_had(monkeypatch):
 
 def test_a_bare_local_id_takes_the_quant_a_plain_load_would(monkeypatch, tmp_path):
     # list_local_gguf_variants orders by descending size, so the head is the biggest
-    # quant. Resolving a bare id to that could evict a working model and then OOM
-    # starting an F16 on a box sized for the Q4 sitting right next to it, and
-    # /v1/models advertised the same head for pinning.
+    # quant. Resolving a bare id to that could evict a working model and then OOM on an
+    # F16 next to a fitting Q4, and /v1/models advertised the same head for pinning.
     from core.inference.local_model_resolver import _local_gguf_entry
 
     for name, size in (("model-F16.gguf", 900), ("model-Q4_K_M.gguf", 100)):
@@ -4263,9 +4260,8 @@ def test_local_and_remote_agree_on_the_preferred_quant():
 
 
 def test_a_just_downloaded_model_is_evidence_before_the_scan_indexes_it(monkeypatch):
-    # Retaining the old index covers what was already known, but nothing covers the
-    # model that just landed until the next scan finishes. A bare request for it in
-    # that window was answered by the unrelated resident model.
+    # The retained index covers what was known, but nothing covers the model that just
+    # landed until the next scan: a bare request for it was answered by the resident one.
     import logging
 
     from hub.services import download_lifecycle
@@ -4312,10 +4308,9 @@ def test_a_just_downloaded_model_is_evidence_before_the_scan_indexes_it(monkeypa
 
 
 def test_a_finished_dataset_is_not_recorded_as_a_local_model(monkeypatch):
-    # finalize_worker_exit is shared with dataset downloads. Noting one as a local
-    # model would refuse a bare /v1 request naming that id while another model is
-    # resident, instead of letting a foreign id fall through, and would kick off a
-    # multi-directory model scan for nothing.
+    # finalize_worker_exit is shared with dataset downloads. Noting one as a local model
+    # would refuse a bare /v1 request naming that id instead of letting a foreign id
+    # fall through, and would kick off a multi-directory scan for nothing.
     import logging
     import time
 
@@ -4359,9 +4354,8 @@ def test_a_finished_dataset_is_not_recorded_as_a_local_model(monkeypatch):
 
 def test_two_local_paths_differing_only_in_case_are_not_the_same_model(monkeypatch):
     # _loaded_satisfies lowercased the request and every backend identifier, so on a
-    # case-sensitive filesystem /srv/models/foo.gguf counted as satisfied by a
-    # resident /srv/models/Foo.gguf and returned before the case-preserving compare
-    # further down ever ran. A repo alias must stay case-insensitive.
+    # case-sensitive filesystem /srv/models/foo.gguf read as satisfied by a resident
+    # /srv/models/Foo.gguf. A repo alias must still stay case-insensitive.
     import os
 
     loaded = _FakeBackend(loaded_id = "/srv/models/Foo.gguf")
diff --git a/studio/backend/tests/test_openai_catalog.py b/studio/backend/tests/test_openai_catalog.py
index 37236fbf66..801d8908cf 100644
--- a/studio/backend/tests/test_openai_catalog.py
+++ b/studio/backend/tests/test_openai_catalog.py
@@ -301,9 +301,9 @@ def test_a_loaded_alias_advertises_the_quant_that_is_actually_loaded(monkeypatch
 
 
 def test_a_nested_model_directory_is_not_the_resident_one(monkeypatch):
-    # Two separately indexed models can nest (/models/A holding A, /models/A/sub/B
-    # holding B). A plain prefix test made loading B mark A resident, so a request for
-    # A was answered with B's weights. The innermost indexed model owns the file.
+    # Two indexed models can nest (/models/A holding A, /models/A/sub/B holding B). A
+    # plain prefix test made loading B mark A resident, so a request for A was answered
+    # with B's weights. The innermost indexed model owns the file.
     outer = _Info("/models/A", "A", model_id = "publisher/A")
     outer.path = "/models/A"
     inner = _Info("/models/A/sub/B", "B", model_id = "publisher/B")
@@ -325,10 +325,9 @@ def test_a_nested_model_directory_is_not_the_resident_one(monkeypatch):
 
 
 def test_a_transformers_model_does_not_mark_a_gguf_alias_loaded(monkeypatch):
-    # Every entry in this loop is advertised as GGUF and carries a GGUF quant. A
-    # Transformers model live from a directory that also holds GGUF exports is not one
-    # of them, and marking the alias loaded had the usage examples pin alias:quant that
-    # nothing can serve while switching is off.
+    # Every entry in this loop is advertised as GGUF with a GGUF quant. A Transformers
+    # model live from a directory that also holds GGUF exports is not one, and marking
+    # the alias loaded had the examples pin a quant nothing can serve with switching off.
     unsloth = _FakeUnsloth()
     unsloth.active_model_name = "/srv/models"
     monkeypatch.setattr(inf, "get_inference_backend", lambda: unsloth)
diff --git a/studio/backend/tests/test_training_worker_flash_attn.py b/studio/backend/tests/test_training_worker_flash_attn.py
index 0c2fdc4628..86511987b1 100644
--- a/studio/backend/tests/test_training_worker_flash_attn.py
+++ b/studio/backend/tests/test_training_worker_flash_attn.py
@@ -212,14 +212,11 @@ def _force_missing_fla_imports(monkeypatch):
 def _pin_fla_model_types(monkeypatch):
     """Pin the auto-discovered FLA allowlist to the Qwen GDN families.
 
-    `_discover_fla_model_types` scans the *installed* transformers for modeling
-    files importing `from fla.`, and `models/qwen3_5/` only exists from
-    transformers 5.x. The backend supports `transformers>=4.51`, so on a 4.x
-    install the gate returns False and every Qwen3.5 assertion below silently
-    passes through a no-op instead of exercising the install path. Pinning keeps
-    these tests hermetic across the whole supported transformers range, the same
-    way test_hook_does_not_install_tilelang_for_model_outside_allowlist pins it
-    against newly added FLA model_types.
+    `_discover_fla_model_types` scans the *installed* transformers, and
+    `models/qwen3_5/` only exists from 5.x. The backend supports
+    `transformers>=4.51`, so on a 4.x install the gate returns False and every
+    Qwen3.5 assertion below silently no-ops. Pinning keeps these tests hermetic
+    across the supported range.
     """
     monkeypatch.setattr(
         worker,
diff --git a/studio/backend/utils/openai_auto_switch_settings.py b/studio/backend/utils/openai_auto_switch_settings.py
index a112391edc..9520846283 100644
--- a/studio/backend/utils/openai_auto_switch_settings.py
+++ b/studio/backend/utils/openai_auto_switch_settings.py
@@ -7,10 +7,9 @@ All off by default so existing API behavior is unchanged:
 - ``openai_api_auto_switch_model``: when on, a ``/v1`` request whose ``model``
   names a downloaded local GGUF different from the loaded one transparently
   loads it before serving (llama-swap-style). Unknown names pass through.
-- ``openai_api_auto_download_model``: when on (and auto-switch is too), a
-  ``/v1`` request naming a GGUF repo that is *not* downloaded starts a
-  background download instead of failing. Gated on auto-switch, which is what
-  serves the model once it lands.
+- ``openai_api_auto_download_model``: when on, a ``/v1`` request naming an
+  undownloaded GGUF repo starts a background download instead of failing.
+  Gated on auto-switch, which is what serves the model once it lands.
 - ``openai_api_auto_unload_idle_seconds``: when > 0, the loaded GGUF is
   unloaded after this many idle seconds to free VRAM. Enabled values have a
   60s floor (0 stays "off"): a tiny TTL tears the model down between turns of
@@ -102,11 +101,8 @@ def get_openai_auto_switch_enabled() -> bool:
 
 
 def get_stored_openai_auto_download_enabled() -> bool:
-    """The persisted auto-download flag, independent of auto-switch.
-
-    The settings UI reads this so toggling auto-switch off displays and
-    round-trips the saved value rather than erasing it.
-    """
+    """The persisted auto-download flag, independent of auto-switch, so the UI
+    round-trips the saved value across an auto-switch toggle instead of erasing it."""
     parsed = _coerce_bool(_cached_setting(OPENAI_AUTO_DOWNLOAD_SETTING_KEY, None))
     return parsed if parsed is not None else DEFAULT_OPENAI_AUTO_DOWNLOAD_ENABLED
 
@@ -114,8 +110,8 @@ def get_stored_openai_auto_download_enabled() -> bool:
 def get_openai_auto_download_enabled() -> bool:
     """Whether a /v1 request may download a GGUF repo it names but doesn't have.
 
-    Gated on auto-switch: auto-switch is what loads the model once it lands, so
-    downloading without it would fetch gigabytes nothing can then serve.
+    Gated on auto-switch: that is what loads the model once it lands, so without
+    it we would fetch gigabytes nothing can serve.
     """
     return get_stored_openai_auto_download_enabled() and get_openai_auto_switch_enabled()
 
diff --git a/studio/frontend/src/features/settings/components/usage-examples.tsx b/studio/frontend/src/features/settings/components/usage-examples.tsx
index 6bb92c5589..846aa0380d 100644
--- a/studio/frontend/src/features/settings/components/usage-examples.tsx
+++ b/studio/frontend/src/features/settings/components/usage-examples.tsx
@@ -334,7 +334,7 @@ const KEY_PLACEHOLDER = "sk-unsloth-YOUR_KEY";
 const USE_TUNNEL_KEY = "unsloth_api_use_tunnel";
 // Slow retry while /v1 has nothing to name: a download or load moves no store state.
 const CATALOG_RETRY_MS = 15000;
-// Slower beat once something is servable: idle unload frees a model without
+// Slower beat once something is servable: an idle unload frees a model without
 // touching the store, so residency is never settled for good.
 const CATALOG_IDLE_MS = 60000;
 
@@ -382,9 +382,8 @@ function useExampleModelName(): string | null {
   const [catalog, setCatalog] = useState<OpenAIModel[] | null>(null);
   // A downloaded but unloaded model is only runnable when switching is on.
   const [autoSwitch, setAutoSwitch] = useState(false);
-  // Idle-unload running on its own (UNSLOTH_MODEL_IDLE_TTL, switching off) still
-  // reloads exactly what it freed on the next request. That restores the stored
-  // checkpoint only, never an arbitrary catalog entry, so it is tracked apart.
+  // Idle-unload on its own (UNSLOTH_MODEL_IDLE_TTL, switching off) reloads exactly
+  // what it freed: the stored checkpoint only, never an arbitrary catalog entry.
   const [idleReload, setIdleReload] = useState(false);
   const usableCheckpoint =
     !!checkpoint && !checkpoint.startsWith("external::") && !looksLikePath(checkpoint);
@@ -396,9 +395,9 @@ function useExampleModelName(): string | null {
     let timeoutId: number | null = null;
 
     const update = () => {
-      // null on failure, never [] or false: a transient error is not evidence that the
-      // server holds nothing, and feeding those negatives in blanked every example while
-      // the model was still servable. Keep the last answer and retry.
+      // null on failure, never [] or false: a transient error is no evidence that the
+      // server holds nothing, and those negatives blanked every example while the
+      // model was still servable. Keep the last answer and retry.
       void Promise.all([
         listOpenAIModels().catch(() => null),
         loadOpenAIAutoSwitchSettings()
@@ -443,10 +442,10 @@ function useExampleModelName(): string | null {
         ? `${pick.id}:${pick.quant}`
         : pick.id;
     };
-    // The store keeps a checkpoint across an idle unload, and across the model
-    // being deleted, so it only names a runnable model while the catalog still
-    // lists it: resident, or downloaded with switching able to reload it. A null
-    // catalog means /v1/models has not answered, which is not evidence against it.
+    // The store keeps a checkpoint across an idle unload and across the model being
+    // deleted, so it only names a runnable model while the catalog still lists it:
+    // resident, or downloaded with switching able to reload it. A null catalog means
+    // /v1/models has not answered, which is not evidence against it.
     const entry = catalog?.find((m) => sameBaseModelId(m.id, checkpoint ?? ""));
     const backed =
       catalog === null || (!!entry && (entry.loaded || autoSwitch || idleReload));
@@ -454,9 +453,9 @@ function useExampleModelName(): string | null {
       if (checkpoint.includes(":")) {
         return checkpoint;
       }
-      // Pin the quant the catalog advertises, not the stored one: membership proves
-      // the repo, and the saved quant can name a file deleted while another quant of
-      // the same repo remains. Fall back to the store only before /v1/models answers.
+      // Pin the quant the catalog advertises, not the stored one: membership proves the
+      // repo, and the saved quant can name a file deleted while another quant remains.
+      // Fall back to the store only before /v1/models answers.
       const quant = catalog === null ? ggufVariant : entry?.quant;
       return quant ? `${checkpoint}:${quant}` : checkpoint;
     }
diff --git a/tests/studio/test_usage_examples_model_source_contract.py b/tests/studio/test_usage_examples_model_source_contract.py
index 446cc619b2..a0c0d04020 100644
--- a/tests/studio/test_usage_examples_model_source_contract.py
+++ b/tests/studio/test_usage_examples_model_source_contract.py
@@ -50,9 +50,9 @@ def test_examples_never_print_a_hardcoded_model_id():
 
 
 def test_catalog_refresh_follows_the_loaded_model():
-    # A dep list that misses these never re-ran, so a finished load left the first
-    # fetch's name. It must not be gated on having no checkpoint either: the store
-    # keeps one across an idle unload, which changes nothing React can see.
+    # A dep list missing these never re-ran, so a finished load left the first fetch's
+    # name. Nor may it be gated on having no checkpoint: the store keeps one across an
+    # idle unload, which changes nothing React can see.
     src = USAGE_EXAMPLES_TSX.read_text(encoding = "utf-8")
     hook = src[src.find("function useExampleModelName") : src.find("// Backend PATH detection")]
     assert "}, [checkpoint, ggufVariant]);" in hook
@@ -66,9 +66,9 @@ def test_catalog_refresh_follows_the_loaded_model():
 
 
 def test_a_stored_checkpoint_needs_catalog_evidence():
-    # The store keeps a checkpoint across an idle unload and across the model being
-    # deleted. Preferring it on the switch setting alone kept naming one /v1/models
-    # had already proved absent, so the snippets 404d instead of falling back.
+    # The store keeps a checkpoint across an idle unload and across a deletion, so
+    # preferring it on the switch setting alone named a model /v1/models had proved
+    # absent, and the snippets 404d instead of falling back.
     src = USAGE_EXAMPLES_TSX.read_text(encoding = "utf-8")
     hook = src[src.find("function useExampleModelName") : src.find("// Backend PATH detection")]
     assert 'const entry = catalog?.find((m) => sameBaseModelId(m.id, checkpoint ?? ""));' in hook
@@ -79,8 +79,8 @@ def test_a_stored_checkpoint_needs_catalog_evidence():
 
 def test_standalone_idle_unload_still_names_the_stored_checkpoint():
     # UNSLOTH_MODEL_IDLE_TTL without auto-switch reloads exactly what it freed, so the
-    # stored checkpoint stays runnable after an idle unload and the panel must keep
-    # showing it. The stash restores only that model, so it can never pick catalog[0].
+    # stored checkpoint stays runnable and the panel must keep showing it. The stash
+    # restores only that model, so it can never pick catalog[0].
     src = USAGE_EXAMPLES_TSX.read_text(encoding = "utf-8")
     hook = src[src.find("function useExampleModelName") : src.find("// Backend PATH detection")]
     assert "const [idleReload, setIdleReload] = useState(false);" in hook
@@ -92,9 +92,9 @@ def test_standalone_idle_unload_still_names_the_stored_checkpoint():
 
 
 def test_a_failed_refresh_does_not_erase_what_the_server_holds():
-    # Catching into [] and false made a transient error authoritative: the panel
-    # dropped a still-servable model and printed "No model" until the next poll.
-    # The catalog is deliberately tri-state, and a failure must stay the unknown one.
+    # Catching into [] and false made a transient error authoritative: the panel dropped
+    # a still-servable model and printed "No model". The catalog is deliberately
+    # tri-state, and a failure must stay the unknown state.
     src = USAGE_EXAMPLES_TSX.read_text(encoding = "utf-8")
     hook = src[src.find("function useExampleModelName") : src.find("// Backend PATH detection")]
     assert "listOpenAIModels().catch(() => null)" in hook
@@ -108,9 +108,9 @@ def test_a_failed_refresh_does_not_erase_what_the_server_holds():
 
 
 def test_the_pinned_quant_comes_from_the_catalog():
-    # Catalog membership proves the repo, not the saved quant. The stored one can
-    # name a file deleted while another quant of the same repo remains, and pinning
-    # it emitted repo:deleted-quant, a missing-quant 404 with a runnable one listed.
+    # Catalog membership proves the repo, not the saved quant: the stored one can name
+    # a file deleted while another quant remains, so pinning it 404d on a missing quant
+    # with a runnable one listed.
     src = USAGE_EXAMPLES_TSX.read_text(encoding = "utf-8")
     hook = src[src.find("function useExampleModelName") : src.find("// Backend PATH detection")]
     assert "const quant = catalog === null ? ggufVariant : entry?.quant;" in hook