Direct access on /model/info was read from the user record alone, so an
unrestricted user calling with a key limited to a few models saw every non-team
deployment, including ones the key gets a 403 on. Resolve the key's grant the
same way and intersect the two.
Resolving a grant now also expands access groups, which the key path needs and
the user path was missing.
Claude-Session: https://claude.ai/code/session_01XL7LBFEew4wi8gphVCDq6n
Team-scoped deployments keep the internal model_name_{team_id}_{uuid} routing key and expose the public name in model_info.team_public_model_name. The dashboard links team model chips with the public name, so the exact filter now matches either name via the existing helper.
The registry key was never copied into ModelInfo, so /v1/model/info reported
null for every model, /model_group/info reported false for every group, and
litellm.supports_parallel_function_calling() returned False for provider-prefixed
entries that declare true. Copy it like every other capability flag and pin the
three surfaces with regression tests.
Resolves LIT-6340
A complexity-router setting placed beside complexity_router_config, or inside a
tier entry's litellm_params, is read by nobody: the router loads its settings only
from litellm_params.complexity_router_config. It does not stay inert. The
alias-marker forwarding and the per-tier param spread carry every unrecognized key
onto the outbound request, and all_litellm_params only knows the outer names, so
the key reaches the provider as an unknown body field and every call through that
model group fails with an error naming an internal config key.
Guard the whole set, derived from ComplexityRouterConfig.model_fields so a field
added later is covered, and scoped to complexity-router deployments because the
names only mean this there (embedding_model is a legitimate flat param on an
s3_vectors vector store). Scope is read from the same merged field view the naming
check is judged on, so a router named only by its default model is in scope and a
field added to the required-field table is covered without another edit. The write
endpoints reject with a 400 naming the keys and where they belong, config.yaml
refuses to start for the same reason max_agentic_loops does, and a tier entry is
judged by the config model itself.
An already-stored deployment keeps loading, so an upgrade cannot take a running
gateway down over a row that was written before the gate existed.
* feat(health): opt-in model-group allowlist for background health checks and health-check routing
* fix(health): merge shared health states per writer scope instead of replacing
* refactor(health): drop restating comment and parameterize test scope annotations
* chore: remove stray generated prisma migration file
* fix(health): merge health states against the Redis snapshot, not the pod-local copy
* fix(health): fall back to the pod-local snapshot when the Redis read returns nothing
reload_search_tools_from_db is a read-modify-write of the shared llm_router
global: it reads the whole table, merges the config tools in, and replaces
router.search_tools wholesale. Two of those interleaving lets the older
snapshot's assignment land last and put back a tool the newer one deleted, so a
revoked tool keeps serving on the provider key it carried until the next reload.
Take MODEL_RECONCILE_LOCK, which add_deployment already uses to serialize the
same shape of work on the same global. It has to go on this entry point rather
than in _init_search_tools_in_db, because _init_non_llm_objects_in_db calls that
while already holding the lock and asyncio.Lock is not reentrant.
A separate search-tools-only lock would not close the race: the periodic
reconcile reaches _init_search_tools_in_db under MODEL_RECONCILE_LOCK, so only
that same lock orders an endpoint refresh against a cron tick.
Ordering across workers is unchanged and still reconciles on the next tick.
Creating a search tool through the UI only wrote the row; the router was updated
solely by the add_deployment job, so the tool was unusable for up to
PROXY_CONFIG_RELOAD_INTERVAL_SECONDS (30s by default) even on the worker that
served the write. Tools declared in config.yaml load straight into the router at
startup, which is why they never showed the delay.
The create, update and delete endpoints now refresh the router inline, matching
what the MCP server endpoints already do. The refresh is best-effort: the row is
already committed, so a failure must not surface as a 500 and push the caller
into a retry that creates duplicates.
Two related gaps go with it. _init_search_tools_in_db skipped the router update
whenever the merged list came back empty, so deleting the last search tool left
it live in memory forever. And in store_model_in_db-off deployments the
add_deployment job is never scheduled, so DB-backed search tools never reached
the router at all; that branch now loads them at startup and keeps them fresh on
its own interval, the same way MCP servers already do.
Adds `general_settings.model_list_healthy_only`, which makes `/models`,
`/v1/models/{id}` and `/model/info` hide models whose backing deployments are
all marked unhealthy by background health checks, for every caller, without
each client having to pass `healthy_only=true`. `/model/info` also gains the
per-request `healthy_only` parameter that `/v1/models` already had.
Everything here is opt-in. With the setting absent, the endpoints take the same
code path they do today and no health lookup runs at all.
The listing filter reads the deployment health cache, which until now was only
populated when `enable_health_check_routing` was on, so `healthy_only=true`
silently did nothing in a plain `background_health_checks` setup. The setting
now also keeps that cache filled. That is a pure write: every routing-time
reader is itself gated on `enable_health_check_routing`, and the cooldown and
failure bookkeeping stays behind that flag, so routing is untouched.
Filtering stays presentation-only and fails open. A hidden model is still
callable, and missing, stale or empty health state hides nothing.
The ceiling used to go through `int(... or 3)`, so anything `int()` accepted
worked. Tightening the new shared validator to `isinstance(int)` turned a
config that boots today into a proxy that refuses to start, because
`max_agentic_loops: os.environ/MAX_AGENTIC_LOOPS` is resolved to a string
before it reaches either check, and a YAML-quoted "5" is a string too.
Accept ints, integral floats, and strings that parse to a whole number. Keep
refusing bools, fractional floats, words, and anything below 1.
The ceiling was only checked at the feature level, on
litellm_settings.websearch_interception_params. The per-deployment
litellm_params.max_agentic_loops, which wins over it, went straight into
int(kwargs.get("max_agentic_loops", 3) or 3), so a 0 was swallowed by the
falsy fallback and read as the default 3. Asking for the tightest ceiling
handed you the loosest one. A non-integer booted the proxy and then failed
every request to that model with "invalid literal for int() with base 10".
Both settings now share one validator, which names the field it rejected,
and the per-deployment value is checked while the model list is read at
startup so a bad value stops the proxy rather than surfacing per request.
The check sits in load_config rather than on LiteLLM_Params because the
proxy builds its router with ignore_invalid_deployments=True, where a
validation error drops the deployment silently instead of refusing to
start. This is the same placement the complexity_router_config plugin
check already uses.
Chat completions read the same key through a separate path that turned 0
into 1 and true into a ceiling of 1, so it now shares the validator too
and the key means one thing on both surfaces.
`pytest.raises(Exception)` with no `match=` passes on any error that broad. A
TypeError from a refactor, a botched fixture, an import that moved: all of them
read as the rejection the test claims to police, so the test goes green for the
wrong reason and stays green after the behaviour it guards is gone.
PT011 closes that gap for the 317 sites B017 could not reach, because B017 only
fires on a single-statement body with no `as e` binding. Each pattern here is the
message the code actually raised, recorded by running the sites under a plugin
that logged the concrete type and text per call site, so the assertions describe
observed behaviour rather than a guess. Where a site raises more than one message
across its parametrize cases, the pattern is an alternation of what was seen;
where the exception carries an empty `str()` and puts the text on `.message`, the
site keeps a narrow `noqa` with the reason.
PT014 removes four parametrize cases that were listed twice. The duplicate re-runs
an assertion that already passed, and it usually marks a case someone meant to
vary and forgot to edit.
* test(lint): ban blind pytest.raises(Exception) with ruff B017
A bare pytest.raises(Exception) accepts whatever the body throws. The TypeError
a refactor introduces satisfies it exactly as well as the rejection the test was
written for, so the crash reads as a pass and the test never goes red.
All 111 existing sites are narrowed here. A runtime probe recorded the concrete
exception each one actually catches, and each site now names that type. Where
the code under test genuinely raises a bare Exception, the site pins a stable
slice of the message with match= instead.
Two sites tell on themselves. The shared responses-API cancel test raises
"custom_llm_provider is required but passed as None" rather than talking to a
provider at all, because cancel_responses takes a provider, not a model. And
test_bedrock_guardrails_with_streaming was the only test in its file still
passing without AWS credentials, because the NoCredentialsError boto3 raised
long before the guardrail ran satisfied the blind raises.
* fix(test): widen the openai batch-dispatch assertion to OpenAIError
The narrowed NotFoundError only holds where OPENAI_API_KEY is set. Without one
the SDK raises OpenAIError while building the client, long before any 404, so CI
went red. OpenAIError covers both and still rejects a TypeError from a refactor.
tiktoken's BPE merge loop is quadratic in the length of a single regex piece, so a long
run of one repeated character turns a multi-MB payload into minutes of CPU. Encoding in
bounded chunks makes that linear, at a drift of at most ~1 token per chunk boundary.
Chunking alone only makes the stall shorter, so the async paths now count in a worker
thread: tiktoken releases the GIL for its Rust encode, so the loop keeps serving other
requests while a count is in flight. The /utils/token_counter endpoint awaits the new
atoken_counter, and the router's async deployment selection counts off-loop and hands
the result to _pre_call_checks instead of making it count inline.
The chunk size knob is bounded to [1, 4096]: a non-positive value used to raise or
silently report zero tokens, and an arbitrarily large one restored the quadratic cost
this exists to remove. Out-of-range and unparseable values warn and fall back to 1024.
Co-authored-by: Yassin Kortam <yassin@berri.ai>
* feat(ui): let admins supply a dark-mode variant of their custom logo
A deployment branded through UI_LOGO_PATH got its light artwork on the
dark sidebar, and there was nothing an admin could set to change that.
Adds UI_LOGO_PATH_DARK, exposed as the logo_url_dark theme setting and a
second field on the UI theme page. /get_image now walks an ordered list
of candidates for the requested theme and serves the first usable one:
the dark logo, then the light logo, then the bundled default.
Falling through rather than failing is the point. An admin who never
sets a dark logo keeps their own light one instead of reverting to
LiteLLM's, and a dark logo that goes missing later degrades to their
light logo rather than dropping their branding entirely.
* fix(ui): recover from a dark logo the browser cannot load
A dark logo given as an http(s) URL is loaded by the browser straight
from the sidebar, so it never passes through the proxy's fallback chain.
A URL that 404s left a broken image where the admin's light logo should
have been, while the same logo given as a local path fell back cleanly.
The sidebar now remembers the dark URL that failed and drops to the light
logo, matching how the proxy resolves an unusable dark logo and how the
provider Logo component already handles a broken image.
The bundled logo is a JPEG, so it carries no alpha and its white
background renders as a bright slab against a dark sidebar. Making it
transparent alone would not be enough either: the wordmark is near-black
and would disappear on dark.
Adds logo_dark.png, derived from the light logo. The sky-blue disc and
train are kept as they are behind a circular alpha mask, and the
wordmark's antialiasing is un-flattened from white into straight alpha
and repainted in the dark theme's own foreground colour. Both files are
1000x257, so swapping between them cannot shift the sidebar header.
/get_image gains a theme query param. The default response is byte for
byte what it was, and a logo configured through UI_LOGO_PATH is served
unchanged in both themes, since custom logos have no dark variant yet.
`ProxyConfig.get_config()` walked the parsed config and replaced every
`os.environ/<KEY>` string with `get_secret(value)` before anything initialized
the secret manager, so a key held only by the manager resolved to `None` and
that `None` was written back into the config. The later fallback in
`load_config` could not recover it, because the key now existed with a `None`
value.
Hoist the initialization into `get_config()`, ahead of the resolution pass, so
every entrypoint gets it: the CLI already did this itself, but the microservice
entrypoints (`gateway/main.py`, `backend/main.py`) uvicorn the app directly and
bypass the CLI. `load_config`'s own call is now redundant and is dropped, so
startup builds the manager once instead of building one and discarding it.
`get_config()` also runs on management-endpoint request paths, so this returns
early once a manager exists rather than rebuilding the client per request.
Also warn when a reference the manager would have been asked for resolves to
`None`. The reporter had no log line at all to work from. `get_secret` only
reaches the manager when reads are enabled and the name is in `hosted_keys`, so
`secret_manager_would_be_consulted` mirrors that gate and keeps the warning off
env-only references, which are expected rather than an error.
Both surfaces wrote zero bytes for the whole time-to-first-token, so an
intermediary with an idle read timeout drops a healthy connection before the
first token. They reached neither keepalive engine, which is what #37322 left
open.
The streaming assistants run spends that wait inside the awaited call that
produces its response, since create_response buffers the first chunk, so it
takes the same open_sse_before_first_byte seam the native routes use. The A2A
route only contacts the upstream agent once its body iterator is first pulled,
so nothing is awaited before the response exists and the gap has to be filled
from inside the stream instead; wrap_sse_stream_with_keepalive_pings already
does that and now takes the filler as a parameter, so A2A gets an SSE comment
its JSON-RPC clients discard rather than Anthropic's ping event.
Off until an operator sets litellm_settings.sse_keepalive_ping_interval_seconds.
* feat(complexity_router): custom classifier plugins via classifier_type 'plugin'
Adds a third classification mode where an operator-supplied hook decides the
tier instead of the heuristic scorer or the LLM classifier. The hook implements
an async classify(context) returning a tier name (built-in value, tier_labels
label, or tier_definitions name) or None to decline; failures, timeouts, and
unknown tiers fall back exactly like a failed LLM classifier. The context
carries the request messages and metadata, including caller identity, so a
plugin can route by team, spend, or any business rule.
The plugin resolves from a dotted path at proxy startup with a load-time check
that classify is a coroutine function, and is closed off over HTTP like the
routing plugins list. Routing decisions record the new classifier_plugin cause.
tier_definitions now accepts classifier_type 'plugin' alongside 'llm'.
* fix(proxy): resolve plugin dotted paths in _delete_deployment before hashing ids
The db-sync reconcile re-reads the raw config and hashes litellm_params to
compute which ids the config wants served, but the router's ids were hashed
from the resolved params where plugin dotted paths are live instances. The
mismatched ids made the reconcile evict every plugin-bearing auto-router one
sync after startup, on any proxy with a database connected. This also affected
the existing routing plugins list, not just the new classifier plugin.
Resolving the plugins in _delete_deployment the same way load_config does makes
both sides hash the same canonical form. A plugin module broken on disk at
reconcile time skips cleanup instead of evicting valid deployments, matching
how a get_config failure is handled
* fix(complexity_router): treat non-string plugin verdicts as declines, centralize the empty-mapping sentinel
A hook returning a non-string raised inside resolve_classified_tier outside the
plugin exception boundary, failing the request instead of falling back. Also
moves the read-only empty mapping to constants.py per repo convention and moves
the classifier plugin product docs out of the package README for the docs repo
* refactor(complexity_router): rename the plugin classifier mode to classifier_type 'custom'
The mode value now names the operator's intent while classifier_plugin keeps
naming the mechanism; routing decisions keep the classifier_plugin cause
* refactor(proxy): pin plugin-bearing deployment ids from the raw params instead of resolving in the reconcile
Replaces the previous approach of re-running plugin resolution inside
_delete_deployment, which imported operator modules on every reconcile cycle
and skipped the whole cleanup pass when any one module was broken on disk.
load_config now stamps model_info.id from the raw litellm_params before
resolution swaps dotted paths for live instances, so the reconcile's raw-config
hash matches by construction and needs no resolution at all: a broken module
cannot stall cleanup for unrelated models, and any future param-transforming
resolution is covered by the same pin. _generate_model_id becomes a staticmethod
so the pin can run before the Router exists; its statically dead non-string key
branches are removed. Also documents candidate_models as an informational
snapshot for classifier plugins, unlike the narrowing surface RoutingPlugin
filters
* fix(router): restore _generate_model_id key handling, align classifier context with the routing-plugin pattern
The staticmethod conversion accidentally dropped the non-string-key branches
from _generate_model_id, a silent hash change for any params with non-string
keys; they are restored verbatim. The classifier plugin context now follows
the Router-level routing-plugin recipe exactly: structured messages come from
resolve_structured_messages over the raw messages, and the metadata key comes
from the shared get_metadata_variable_name_from_kwargs helper, which also
replaces the duplicated inline sniff in _pick_model_for_tier. This removes the
raw-or-resolved fallback where a plugin could silently receive resolved
messages when a call site forgot to pass the raw ones
* refactor(router): make generate_model_id public, guard classifier context construction
Two modules legitimately hash deployment ids with the same helper now (Router
and the proxy's config-load pin), so the private name was lying about its
audience and the cross-module call needed a pyright suppression; renaming it
public restores the static safety net. The classifier plugin's RoutingContext
construction moves inside the failure boundary, matching the LLM path where
litellm-side prompt building also falls back rather than failing the request,
and a prompt-only call with no message list is now covered by a test
The Global Control Plane (formerly documented as the HA Control Plane) is
documented as an Enterprise feature, but `worker_registry` carried no premium
check, so any OSS install could run one. Gate it at config load, matching the
`enforced_params` precedent, and fail startup rather than ignoring the registry
silently: a silently dropped registry degrades a control plane into an ordinary
proxy with no signal to the operator.
Also declare `worker_registry` and `general_settings.control_plane_url`, both
load bearing today and neither previously declared, so they appear in the
generated config schema.
Anthropic's Models API declares max_input_tokens and max_tokens as nullable, not
optional, and the live vendor endpoint returns both keys on every entry. The
merged Anthropic-native listing dropped either key whenever LiteLLM could not
resolve a limit, so a client validating against a nullable-but-required schema
saw a malformed entry for any model the cost map does not know.
* feat(proxy): serve Anthropic-native /v1/models for Claude Code gateway discovery
* refactor(proxy): move Anthropic model-list formatter into llms/anthropic/common_utils
* fix(proxy): make model_list request param optional for direct callers
* style: apply ruff format to changed lines
* style: satisfy ruff strict-rule budget (UP006, I001)
* style: satisfy type-discipline budget (LIT002 mutable-ok, LIT009 pyright ignore)
* style: satisfy LIT001/LIT010 and drop explanatory comment per contributor rules
* fix(proxy): translate team model names in the Anthropic /v1/models response
* ci: trigger buildkite status report
* feat(proxy): carry token limits into the Anthropic-native /v1/models entries
* fix(proxy): cast the injected request so the anthropic-version guard is a real comparison
* fix(proxy): explain the model listing casts so the type-discipline gate passes
---------
Co-authored-by: yuneng-jiang <yuneng@berri.ai>
Co-authored-by: Yassin Kortam <yassin@berri.ai>
* fix(proxy): expand config-defined model access groups when resolving team models for /v2/model/info
Teams whose only model grant is a config-defined access group (a model_info.access_groups
name listed in team.models) got an empty /v2/model/info?include_team_models=true result.
_add_team_models_to_all_models passed each team.models entry straight to
llm_router.get_model_list(model_name=...), which never matches an access-group name, so
the group's member deployments were dropped. Runtime auth and /v1/models were unaffected
because they expand team.models through get_team_models first.
Resolve team.models through the same get_team_models resolver before iterating, reusing the
exact path runtime auth and /v1/models trust so the two can't drift again. The get_model_names
and get_model_access_groups accessors are hoisted above the team loop so they run once.
* fix(proxy): keep a literal model whose name collides with an access-group name in listings
A grant string that names both a deployed model and a config access group grants
BOTH at runtime (_check_model_access_helper unions them), but the listing resolver
dropped the literal and substituted the group members, hiding a callable model from
/v1/models and /v2/model/info. Keep the literal when it is also a deployed model so
listings match runtime access exactly. Pure-group names (no collision) are still
replaced by their members. Also rewrites _get_models_from_access_groups to build
its result without mutating the input list.
Addresses the Greptile P1 on this PR.
* fix(proxy): type proxy_model_list param as Sequence to satisfy LIT001 budget
The outer wrap_sse_stream_with_keepalive_pings layer duplicated the
keepalive engine that PR #34423 already runs inside async_data_generator
for chat completions and responses streams, and it kept pinging
deployments whose operator set keepalive_seconds: 0 as a hard disable.
sse_keepalive_ping_interval_seconds is now the global fallback inside
_resolve_keepalive_seconds, so deployment and request values keep
precedence, an explicit 0 still disables, the [1, 300]s clamp applies,
and router-less proxies arm the wrap when the global default is set.
* fix(alerting): dedupe scheduled Slack spend reports across pods
Every pod ran its own weekly/monthly spend report jobs, prometheus
fallback stats cron, and daily report loop, so deployments with
multiple replicas or uvicorn workers received one copy per pod.
Gate each scheduled send behind the shared PodLockManager redis lock.
The lock is never released: its TTL (the full reporting window for the
weekly interval job, whose per-pod anchors drift by boot time and
jitter) doubles as a sent-this-window marker. acquire_lock returning
None (no redis wired) proceeds, preserving single-pod behavior.
Also generalize the pod lock could-not-acquire log line, which claimed
to be about spend tracking for every consumer.
Fixes#14809
* fix(alerting): harden spend report locks after adversarial review
Weekly lock TTL gets an hour haircut: with ttl equal to the interval,
the winner re-fires just before its own key expires, reacquires without
a TTL refresh, and the key then lapses in time for a trailing pod to
re-send. Job/lock ids move to litellm/constants.py per convention, and
spend_report_frequency now rejects non-positive day counts, which
previously coerced to an every-second schedule and would now compute a
negative lock TTL that silently never sends.
Adds the missing test coverage the review flagged: startup_event's
pod_lock_manager wiring (identity-asserted), the prometheus closure's
positive path, and the ungated immediate prometheus send pinned to
exactly one await.
* test(alerting): consolidate spend_report_frequency validator coverage
Drops a duplicate non-positive-days test and parametrizes the survivor
over the suffix half of the validator too
* fix(alerting): route the startup prometheus fallback send through the pod lock
Greptile caught that the boot-time send still ran once per pod when
PROMETHEUS_URL is set, the same duplication class this PR removes
* fix(alerting): make report lock acquisition non-reentrant
Greptile caught that a pod booting within an hour of the fallback stats
cron sent twice: the startup send takes the lock, then the cron fire
hits acquire_lock's reacquire branch, which returns True for the
holder. Window-marker gates now pass allow_reentrant=False so a live
lock blocks everyone including its holder; leader-election consumers
keep the reentrant default
* test(proxy): give spec'd ProxyLogging mocks a db_spend_update_writer
_initialize_slack_alerting_jobs now reads it for the pod lock manager,
and spec=ProxyLogging blocks instance-only attributes
* feat(proxy): add per-deployment keepalive_seconds SSE heartbeat for long-running streams
Adds _iter_with_keepalive, _keepalive_from_deployment_config, and
_resolve_keepalive_seconds helpers to proxy_server.py. When enabled
(keepalive_seconds > 0 in request body or deployment litellm_params),
async_data_generator emits ': ping\n\n' SSE comment frames every N
seconds during idle upstream intervals, preventing load-balancer
idle-timeout drops on long chain-of-thought reasoning streams.
The hot path (keepalive_seconds absent or 0) is a plain async-for with
no per-chunk Task wrapping — zero overhead. Includes 8 new unit tests
covering sentinel emission, hot-path pass-through, early-close cleanup,
priority resolution, deployment-config lookup, and end-to-end heartbeat
emission through async_data_generator.
Registers keepalive_seconds in all_litellm_params (types/utils.py) so
the parameter is not stripped from request bodies. Adds the field to
LiteLLMParamsTypedDict and GenericLiteLLMParams (types/router.py) so
deployment YAML config is parsed and validated.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(proxy): narrow BaseException to CancelledError to fix BLE001 strict lint gate
* fix: use explicit None check instead of truthiness in keepalive_seconds extraction
`float(raw or 0)` would treat any falsy value (including the integer 0)
as absent and substitute 0.0 before float() saw it. Replace with
`float(raw) if raw is not None else 0.0` so a caller-supplied zero is
correctly passed through to the `value <= 0` guard that disables
keepalive, rather than being silently overwritten.
* fix(proxy): don't guess a deployment's keepalive_seconds when model_id is missing
When a streaming response lacks _hidden_params.model_id, the fallback that
looks up keepalive_seconds by model_name previously returned the first
configured deployment's value, which could apply the wrong interval (or
override an explicit disable) when multiple deployments share the same
model_name with different keepalive_seconds settings. Only resolve the
fallback when every deployment agrees; otherwise leave it unset.
* fix(proxy): also treat an unset keepalive_seconds as disagreement in the fallback
The model_name fallback for keepalive_seconds only compared configured
values, filtering out deployments that leave the field unset entirely.
That meant a deployment with no keepalive_seconds configured could still
inherit a sibling deployment's interval when model_id is unavailable.
Compare the raw per-deployment value (including None for unset) so an
unconfigured deployment never silently adopts another's heartbeat.
* fix(proxy): deployment-level keepalive_seconds: 0 is a hard disable clients can't override
Previously an authenticated client's request-level keepalive_seconds always
took precedence over the deployment default, including when a deployment
operator explicitly set keepalive_seconds: 0 to disable heartbeats. That let
any client re-enable heartbeats for a deployment the operator opted out of,
using them to keep an idle-looking stream alive past a load balancer's idle
timeout and hold a parallel-request slot open longer than intended.
Treat an explicit deployment-level 0 as authoritative: resolve the
deployment's configured value first, and short-circuit to disabled before
ever looking at the request body if the deployment hard-disabled it.
* fix(proxy): a stale (unresolvable) model_id must not fall through to model_name guessing
A populated _hidden_params.model_id names the specific deployment that
served a stream. If that ID no longer resolves (e.g. a deployment removed
by a config reload mid-stream), the resolver was falling through to the
model_name-based fallback, letting a currently-live sibling deployment's
keepalive_seconds silently apply to a stream it never served. Return None
once a populated model_id fails to resolve, rather than degrading to a
guess.
* fix(proxy): keepalive_seconds is operator-only by default; require deployment opt-in for client override
A security review flagged that a client's request-level keepalive_seconds
could unilaterally enable heartbeats for any deployment, even one that
never configured keepalive_seconds at all, letting an authenticated client
defeat load-balancer idle timeouts and hold a parallel-request slot open
for longer than the deployment operator ever intended, with no way for
the operator to prevent it short of explicitly setting keepalive_seconds: 0.
Add allow_client_keepalive_override (default False) to LiteLLMParamsTypedDict
and GenericLiteLLMParams. _resolve_keepalive_seconds now ignores the request
body's keepalive_seconds entirely unless the resolved deployment explicitly
grants override permission; only the deployment's own configured value (or
disabled, if unset) applies otherwise. An explicit deployment-level 0 still
takes priority over everything, including a grant of override permission.
* fix(proxy): register allow_client_keepalive_override in all_litellm_params
Caught during live proxy verification against the real Anthropic API:
allow_client_keepalive_override was added to LiteLLMParamsTypedDict and
GenericLiteLLMParams but never registered in all_litellm_params, so it
leaked straight through into the provider request body as an unrecognized
field. Anthropic rejected every call on a deployment that had this field
configured with a 400 ("Extra inputs are not permitted"), regardless of
its value. Register it alongside keepalive_seconds so it's stripped
before reaching the provider, matching what keepalive_seconds already
does.
* feat(proxy): support keepalive_seconds via x-litellm-keepalive-seconds header
Some clients (e.g. the Vercel AI SDK) can set custom headers more easily
than extra JSON body fields. Add x-litellm-keepalive-seconds, following
the existing x-litellm-timeout/x-litellm-stream-timeout/x-litellm-num-retries
convention in LiteLLMProxyRequestSetup: the header merges into the same
data["keepalive_seconds"] field the request body already populates, so it
goes through the exact same _resolve_keepalive_seconds precedence and the
allow_client_keepalive_override gate -- a header can't enable heartbeats
for a deployment that hasn't opted in any more than the body field can.
Verified live against the real Anthropic API: the header produces real
heartbeats on an opt-in deployment (88 pings over a genuine long-reasoning
stall) and is silently ignored on a deployment without override permission
(0 pings), matching the existing body-field behavior exactly.
* chore: rebase onto litellm_internal_staging, drop unrelated credential_migration.py reformat, fix budget-ratchet drift
Rebased onto the current litellm_internal_staging (merge-base was 5 days
stale). Dropped the now-redundant schema.d.ts-only regen commit entirely
(the new base's own schema.d.ts already supersedes it) and regenerated
schema.d.ts fresh against the new base.
Reverted litellm/proxy/management_endpoints/credential_migration.py to
exactly match litellm_internal_staging: it was a pure reformat with no
semantic change, unrelated to this PR, flagged by review as unnecessary
noise in an encryption-migration file.
Fixed two lint-budget-ratchet failures caused by the base's ceilings
tightening since this branch last synced (other merged work lowered
ANN401/LIT001 budgets; this code was previously under budget and didn't
change):
- _iter_with_keepalive's aiter param: Any -> AsyncIterator[Any], a real
narrowing (it's always the result of .__aiter__()).
- _keepalive_from_deployment_config/_resolve_keepalive_seconds's
request_data param: dict[str, Any] -> Mapping[str, Any], matching the
existing read-only-dict convention already used elsewhere in this file
(_apply_ssrf_general_settings, _build_redis_usage_cache, etc.) for
params that are only ever read, never mutated.
- response/raw params: dropped the explicit `Any` annotation to match
async_data_generator's own (deliberately unannotated) `response` param,
its actual caller.
- litellm_pre_call_utils.py's new headers param: dict -> Mapping[str, str],
same read-only-dict rationale.
* fix(proxy): freeze the transient collections in the keepalive helpers
_iter_with_keepalive and _keepalive_from_deployment_config built a set
literal for asyncio.wait, a set comprehension for the per-deployment
config-agreement check, and two dict-literal fallbacks, all flagged by
the LIT002 mutable-collection-construction gate. Switched to a tuple
for asyncio.wait, a frozenset-wrapped generator plus next(iter(...))
for the config check, and a shared MappingProxyType({}) empty mapping
for the fallbacks.
* fix(proxy): trust metadata.model_info.id over the stale model group after a router fallback
Greptile P1: when a streaming request falls back from model group A to
group B and the response's _hidden_params carries no model_id,
_keepalive_from_deployment_config fell straight through to guessing
via request_data["model"], which still names the pre-fallback group A
since the fallback handler mutates its own local **kwargs copy, not
this dict. request_data[metadata|litellm_metadata]["model_info"]["id"],
by contrast, is mutated on this same dict by
Router._update_kwargs_with_deployment on every attempt including
fallbacks (the same source ProxyLogging._build_litellm_call_info uses
for logging), so check it before falling through to the model-name
guess.
Added two regression tests that fail on the prior code (assert
get_model_list is never called once metadata.model_info.id resolves)
and pass with the fix.
* Revert "fix(proxy): trust metadata.model_info.id over the stale model group after a router fallback"
This reverts commit d7790678645695b20f25880315238b49c31a9143.
* fix(proxy): satisfy the new LIT010/ANN001 gates in the keepalive helpers
litellm_internal_staging picked up a LIT010 (every local/module variable
must be declared Final unless it's genuinely rebound) and tightened
ANN001 (missing parameter annotations) since this branch last synced.
Annotated every single-assignment local and module constant with
Final, suppressed pending's loop-carried reassignment with
# rebind-ok, and typed the previously-bare response/raw parameters as
object with isinstance narrowing at their use sites instead of cast
(LIT006 discourages cast; validate into a concrete type instead).
Also swapped the hand-rolled getattr(response, "_hidden_params", None)
+ isinstance(hidden, dict) check for the existing
get_hidden_params_dict() helper already used for this exact purpose
elsewhere in this file and in common_request_processing.py.
* fix(proxy): re-resolve keepalive_seconds per chunk to track mid-stream fallback
Greptile P1: the router can perform a mid-stream fallback to a
different deployment partway through a stream (MidStreamFallbackError
in router.py), and Router._apply_fallback_hidden_params_to_item merges
the fallback deployment's hidden params onto every subsequent chunk.
But _resolve_keepalive_seconds was only ever called once, before
iteration started, against the pre-fallback response wrapper, so a
stream that fell back to a deployment with a different (or disabled)
keepalive policy kept using the original deployment's interval for the
rest of the stream.
_iter_with_keepalive now takes a resolve_keepalive_seconds(item)
callback and re-resolves after every real chunk using that chunk's own
_hidden_params (which do carry the fallback deployment's identity),
rather than trusting the value picked before iteration began. Updated
the three existing timing tests to inject a constant-returning
resolver, since they pin the sentinel/cancellation mechanics rather
than re-resolution, and added two regression tests (interval lowered
and raised mid-stream) that fail against the prior static-resolve
signature and pass with the fix.
* fix(proxy): keep re-resolving keepalive even when a stream starts disabled
Greptile P1: a stream that starts on a deployment with keepalive off
(or unset) skipped _iter_with_keepalive entirely at the call site, so
a mid-stream fallback to a deployment that enables it never got a
chance to activate heartbeats for the rest of that stream, risking the
exact load-balancer idle-timeout this feature exists to prevent.
_iter_with_keepalive now has an internal fast path for
keepalive_seconds <= 0 that still re-resolves after every chunk (no
asyncio.create_task/wait overhead while inactive, same cost as a bare
async for), so activation from a disabled start works the same way
deactivation and interval changes already do. The caller now only
skips wrapping entirely when there's no router to ever fall back
through in the first place (llm_router is None), rather than whenever
the first chunk's deployment happens to start with keepalive off.
Added a regression test that starts keepalive_seconds=0, has the
resolver enable a short interval on a later chunk, and asserts
sentinels appear afterward; it fails against the prior
call-site-gated code and passes with the fix.
* perf(proxy): memoize keepalive resolution per chunk's model_id
_resolve_keepalive_seconds ran a full llm_router.get_deployment() Pydantic
rebuild after every streamed chunk, even when keepalive was unconfigured
anywhere in the deployment list, since async_data_generator wraps every
stream once a router exists. Caching the result by model_id keeps mid-stream
fallback re-resolution correct while paying the router lookup once per
deployment instead of once per token.
* fix(proxy): expire cached keepalive resolution after a bounded TTL
veria-ai flagged that caching by model_id alone lets an already-in-flight
stream keep evading a live config reload (deployment removed, keepalive
disabled, or client override revoked) for the rest of the stream. Expiring
the memo after _KEEPALIVE_CACHE_TTL_SECONDS bounds that window instead of
freezing the resolved value for the stream's full lifetime, while still
avoiding a full deployment rebuild on every chunk in the steady state.
Also fixes add_litellm_data_for_backend_llm_call's now-required request_data
kwarg in the header-merge test, picked up by rebasing onto
litellm_internal_staging.
---------
Co-authored-by: Deepanshu <deepanshu.lulla@alpha-sense.com>
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(proxy): isolate guardrail load failures per row
One DB guardrail row that fails to initialize aborted the whole
_init_guardrails_in_db loop, so a single typo'd guardrail type or a
missing required param left the proxy running with zero DB guardrails
registered and requests that should have been blocked reaching the
provider.
Catch per row around sync_guardrail_from_db, log the guardrail name, id
and error, and continue with the remaining rows. The failing row's id is
still added to db_guardrail_ids before the attempt so reconcile_db_guardrails
cannot mistake a live row for a deleted one.
* test(proxy): drop inline note and record reconcile via a handler double
Replaces the patched bound method with an InMemoryGuardrailHandler subclass
that records what reconcile_db_guardrails received, so the test injects a
double instead of swapping a method on a live object.
PR #36020 changed AgentRegistry.get_agent_list to return tuple[AgentResponse, ...], and PR #35163 added a test asserting the result equals []. Both were green on their own branches and only collided once they were both on litellm_internal_staging, so proxy-server has been failing on every PR since with 'assert () == []'.
Nothing user-facing was wrong: get_agents only iterates the result and rebuilds it with comprehensions, and FastAPI serializes a tuple to the same JSON array. The test expectation was simply stale, so it now compares against ().
The get_agents local was still annotated list[AgentResponse] while two branches assign the registry tuple straight into it, so it widens to Sequence[AgentResponse]. That covers both the tuple and the list branches without pretending the value is mutable.
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* warn at startup when a proxy-wide budget is set but no DB is connected
litellm.max_budget is only enforced via DB-loaded global spend, so a DB-less proxy silently ignores it. Log a one-time startup warning.
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* refactor(proxy): inject max_budget into DB-less budget warning
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* test(proxy): cover DB-less budget warning startup call
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* test(proxy): pin DB-less budget warning call site
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* test(proxy): stabilize budget warning call-site pin
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
---------
Co-authored-by: tin <tin@berri.ai>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
SGR has had two independent definitions. The admin UI derived it from
SpendLogs, so it counted what litellm's logging callbacks observed and could
attribute and price. BillableRequestMetricsMiddleware counted what the proxy
actually answered at the ASGI edge, but only exported to OTLP for enterprise
metering. The two disagree by design in places, and the SpendLogs figure goes
quiet whenever spend logging is disabled or the callbacks are bypassed.
This adds LiteLLM_DailyGatewayRequests, written by the middleware, and points
the dashboard's Successful Requests tile at it.
Requests fold into an in-memory map at record time rather than going through a
queue like the spend path. A count is a pure aggregate, and every dimension of
the key is chosen by the proxy from a closed set: the date, the category, and a
route that the classifier maps to one of a fixed list of strings rather than
passing the raw path through. Nothing a caller sends can add a key, so the fold
and the table are bounded by (days x categories x routes) however much traffic
arrives; the spend queue blocks once full, which is not acceptable in the
response path. A scheduler job drains it on the existing batch interval, and a
failed flush merges its counts back so a database blip undercounts nothing.
The middleware previously returned early when no billing recorder was
injected, which is the unlicensed case. The new sink is not license-gated, so
that early return now requires both sinks to be absent. The billing recorder
keeps its 2xx-only gate; the sink takes every status so failed_requests is
real. The sink is not told which deployment served the request, unlike the
billing recorder. That id is a sha256 over litellm_params, credentials
included, so a caller who puts a credential in the request body mints a fresh
one per distinct value. No configuration is needed for that: api_base and
base_url are on _BANNED_REQUEST_BODY_PARAMS and need allow_client_side_
credentials, but api_key is not on that list, and both reach the same
_handle_clientside_credential branch. The read endpoint aggregates the
dimension away regardless, so the key is better off without it.
The new table carries no key, user or team dimension, so /gateway/daily/activity
is restricted to proxy admin roles and the per-key and per-model breakdowns
keep reading the daily spend tables. The old path is left running and marked
with TODOs.
A fetched result carries the range key it was fetched for, and the render
selects it only when that key matches the range on screen. Both the gateway
counts and the spend aggregate go through that rule: the request tiles read the
first and fall through to the second, so stamping only one of them would leave
the tile showing a superseded range by the other route.
The paginated pages behind that aggregate are reached through a failure flag,
so the flag is stamped too. A flag left over from the previous range would let
those pages through while a new range is in flight, which is the same defect
one fallback further down.
* fix(proxy): give proxy_admin_viewer read parity with proxy_admin
Route-level checks already default-allow management GETs for the viewer
role, but ~15 handlers compared user_role to PROXY_ADMIN only, dropping
viewers into regular-user scoping (/key/list, /user/info, /model/info,
guardrails, prompts, agents, memory, workflows, MCP catalog, coordination
redis settings, credential migration check, enterprise projects). Swap
those read paths to user_api_key_has_admin_view; write gates unchanged.
The dashboard now presents the viewer session as Admin for all gating
(effectiveSessionRole) so every page fetches with admin visibility, with
userRoleLabel/isViewOnly preserving the account-menu label and the
playground cost guard. The server remains the write authority.
* refactor(agents): remove side-effectful health_check param from GET /v1/agents
Addresses a security review finding on the admin viewer read parity change:
listing agents with health_check=true made the proxy issue a server-side GET
to every agent URL, so a read-scoped caller could trigger request fan-out
beyond their object permissions. The list endpoint is now a pure read for
every role.
Removes the query param, the URL probing helper and its timeouts, the
AgentHealthCheck httpx provider tag, and the dashboard's Health Check
toggle. Requests still passing health_check=true get the full list back
with the param ignored.
* fix(proxy): keep credential encryption check proxy_admin only
The residual scan behind GET /credentials/migrate-encryption/check loads
every model, credential, MCP, team, and verification-token row and runs a
decryption attempt on each stored value. Extending it to proxy_admin_viewer
let a read-only account repeatedly trigger deployment-wide scans, so the
route keeps its original full-admin gate.
* fix(agents): restore health_check, keep list fast path proxy_admin only
Restores the agent health_check feature exactly as before this PR: the
query param, the URL probing helper, the httpx provider tag, and the
dashboard toggle all return, so existing callers keep the filtering
contract. The viewer expansion is instead reverted at its source: the
GET /v1/agents admin fast path stays PROXY_ADMIN only, so a
proxy_admin_viewer goes through the object-permission scoped branch as
before and cannot fan out health checks beyond their allowlist. The
viewer read of a single agent stays viewer-inclusive since it has no
side effects.
Per-window budget enforcement aggregated LiteLLM_SpendLogs on every cold
counter and on every 5s authoritative floor check. SpendLogs has no index on
api_key or team_id, so each check range-scanned the highest-volume table.
Reads now hit the LiteLLM_BudgetWindowSpend row by primary key and only fall
back to the aggregate when no row exists for the window being enforced. A row
is current when its window_start is at or past the caller's expected start, so
a pod holding a stale reset_at trusts a window another pod already rolled
instead of summing the previous window back in.
window_duration is threaded from the budget_limits entry through to the read
rather than parsed back out of the counter key. The reader never writes rows.
* fix(proxy): persist periodic reload schedule state so status survives restarts and fires without store_model_in_db
The model cost map and Anthropic beta headers reload schedules kept their
last-run time in a per-pod module global, so GET /schedule/*/status reported
last_run null after any restart and the Admin UI showed the reload as never
having run. The reload check also only ran from the add_deployment job, which
is registered only when store_model_in_db is true, so config-file deployments
stored a schedule that never fired.
Persist last_run_at and reload_requested_at as dedicated columns on
LiteLLM_Config, owned by the reload job and manual reload endpoints, while the
schedule endpoints own the param_value JSON (interval_hours); no writer can
clobber another's fields. Serve status entirely from the row. Register the
check as its own periodic_reload_job outside the store_model_in_db gate.
Replace the force_reload boolean with a reload_requested_at timestamp each pod
compares against its own in-memory last reload, so a manual reload reaches
every pod exactly once instead of being cleared by the first poller. Run the
blocking fetches via asyncio.to_thread, and stamp last_run_at with update_many
so a schedule cancelled mid-poll is not resurrected.
* fix(proxy): compare reload requests against pod data age seeded at boot
A pod that had never reloaded kept its in-memory clock at None, and with no
interval configured nothing ever set it, so every manual reload request was
ignored by every pod except the one serving the click (Greptile P1 on the
previous commit). Seed the per-pod timestamp at boot as the time its data was
loaded and reload whenever a request or the interval is older than that, which
also removes both None special cases from the due predicate. A schedule whose
row has no last_run_at fires on the next tick so the first run does not wait a
full interval.
* fix(proxy): scope reload persistence to the model cost map and seed the pod clock from the actual load time
Revert the Anthropic beta headers reload path to its previous JSON-flag
implementation so this PR only changes the price data reload; the beta headers
path keeps working exactly as before and can migrate to the shared module in a
follow-up. The unused columns on its config row are inert.
Seed model_cost_map_loaded_at from the timestamp get_model_cost_map records at
the actual import-time fetch instead of ProxyConfig construction time, closing
the startup window where a manual reload request stamped between the fetch and
the constructor compared as older than the pod's data and was skipped
(Greptile P1 on the previous commit).
* refactor(proxy): drop the legacy force_reload backfill from the reload tracking migration
The backfill only carried over a manual reload clicked in the seconds before an
upgrade, and every upgrade restarts the pods, which re-fetch the cost map at
import and so already deliver what that request asked for. Removing it makes
the migration schema-only, so prisma db push and prisma migrate deploy leave
the database in the same state instead of diverging on a data statement that
only one of them runs.
* fix(proxy): stamp reload timestamps at the precision they are stored at
Postgres stores these columns as TIMESTAMP(3) while Python stamps microseconds,
so a pod comparing its in-memory clock against the persisted copy of the same
instant read as newer and skipped the reload request it had just recorded.
Truncate every stamp to milliseconds at the source, and floor the boot seed the
same way, so the in-memory value and its persisted copy compare exactly.
* fix(proxy): identify manual reloads by revision instead of comparing timestamps
Comparing a request timestamp against each pod's data age made correctness depend
on clock resolution: Postgres stores TIMESTAMP(3) while Python stamps microseconds,
and two events inside the same millisecond are indistinguishable no matter how the
comparison is written.
Replace reload_requested_at with a reload_revision counter the manual reload
endpoint increments atomically in the database. Each pod records the revision it
last applied and reloads whenever the row's differs, so a request reaches every pod
exactly once regardless of clock skew or precision, and concurrent requests publish
distinct revisions instead of overwriting one another. A pod adopts the current
revision on its first poll, since data it loaded at boot already satisfies any
earlier request. Interval reloads still key off the pod's own data age, where hour
scale comparisons make precision irrelevant.
* fix(proxy): seed the applied reload revision at startup
A pod adopted whatever revision it found on its first poll, so a manual reload
published while the pod was starting was marked applied without ever being
served and the pod kept the prices it fetched at import. Read the row once at
startup instead, right after that fetch, and treat a missing row as revision 0
* style(tests): revert incidental reformatting of test_proxy_server.py
An earlier ruff format run reflowed the whole file from its 88-column
formatting, adding ~1150 lines of churn unrelated to this PR. Replay only
the real test changes onto the original formatting
* fix(proxy): serve an outstanding reload request on a booting pod
Seeding the applied revision at startup left a window: a manual reload
published after the import-time cost map fetch but before startup read the
row was marked applied without ever being fetched, stranding that pod on
stale prices when no interval was configured. A pod now starts unapplied and
serves any outstanding request on its first poll, which costs one redundant
fetch per boot and removes the window along with the seeding step
* fix(proxy): accept a reload interval still encoded as JSON text
param_value is written with safe_dumps, and a raw row read can return it
decoded or as a string depending on the driver. Strict validation rejected
the string, so the schedule read as disabled and an admin's configured
reloads silently stopped. Mirrors the guard ConfigRepository.get_param
already carries for the same column
* fix(proxy): cancel a reload schedule without resetting the revision
* fix(proxy): null the interval in JSON so cancelling keeps the revision
prisma rejects a null literal for a Json? column, so update_many writes an
interval-less object instead. The fake config table now rejects the same input
the database does, which is what the live run caught and the mock did not.
Also records the run before adopting the revision, so a failed status write
leaves the request unserved for the next poll rather than reporting a run that
never landed.
* fix(ui): match the CI-generated user_role union order in schema.d.ts
* fix(proxy): propagate user_email and bind api_key on JWT auth paths
Standard JWT auth built UserAPIKeyAuth with user_id but never user_email, and the first auto-registered request early-returned a key with token set but api_key unset, so spend-log attribution logged user_api_key_user_email and user_api_key_hash as null. Bind api_key to the token hash on the auto-registered key, copy user_email from the resolved user object on both the standard and auto-register JWT paths, and warn when enable_jwt_auth/litellm_jwtauth are placed at the config top level where they are silently ignored.
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* test(proxy): cover misplaced top-level JWT config warning
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
---------
Co-authored-by: shivam <shivam@berri.ai>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Co-authored-by: ryan <ryan@berri.ai>
* fix(proxy): retry model cost map fetch with Retry-After-aware backoff and stop downgrading to the packaged backup on reload failure
A 429 or transient network error during a manual or scheduled model cost map
reload used to silently replace litellm.model_cost with the stale backup JSON
bundled in the installed wheel, stamp the reload as successful, and clear the
force_reload flag, so a fleet could serve months-old pricing until the next
interval. Runtime reloads now go through refetch_model_cost_map, which retries
429/5xx/transport errors up to 3 times honoring Retry-After (capped at 30s,
exponential backoff with jitter otherwise) and returns a failure value instead
of the backup when the fetch or integrity validation fails. On failure the pod
keeps its currently loaded map, the periodic job leaves last_run and
force_reload untouched so it retries on the next config poll, and the manual
endpoint returns 502 with the reason instead of reporting a fake success.
Startup behavior is unchanged: boot still falls back to the packaged backup
since there is no previously loaded map to keep.
* fix(proxy): use shared async httpx client for cost map reload and make retry tests CI-env-proof
The reload fetch now goes through get_async_httpx_client with a dedicated
httpxSpecialProvider.ModelCostMap pool instead of constructing a raw
httpx.AsyncClient, so it inherits deployment-level TLS and transport settings
and passes the ensure_async_clients gate. Tests inject a MockTransport-backed
client through the same seam. An autouse fixture clears
LITELLM_LOCAL_MODEL_COST_MAP, which CI exports and which short-circuited the
retry tests; the two TestPriceDataReloadAPI tests and the config sync pubsub
reload test that still patched get_model_cost_map now patch
refetch_model_cost_map instead.
Key and team `router_settings.model_group_alias` was accepted, persisted and
echoed back by `/key/info`, but never applied at request time, so the request
ran on the group the caller asked for. `route_request` forwards only the
settings the Router accepts as per-request kwargs, and `model_group_alias` is
not one of them: the Router resolves aliases from its own instance attribute,
which holds the global config map and is shared across requests.
Resolve the alias in the proxy instead, alongside the existing model-alias
rewrites and ahead of the pre-call hooks, so per-model limits and guardrails
key off the group that actually serves the request. Authorize the alias target
before the rewrite; model access was checked against the requested group, so a
key whose alias points at a group it cannot call gets the usual 403 rather than
being quietly served it.
Resolves LIT-4879
