- sync llm_passthrough_route: read and close an error-status streaming
response before mapping it, so upstream 4xx/5xx surface as the provider
error instead of httpx.ResponseNotRead
- AsyncPassthroughStreamingResponse: expose aiter_bytes() and carry
_hidden_params so the router attaches headers in place instead of
wrapping the stream in HiddenParamsAsyncIteratorWrapper, which 500'd
every streaming azure router-model passthrough request
- logging: swap the passthrough httpx result for the transformed
ModelResponse/EmbeddingResponse when firing success callbacks
- get_llm_provider: resolve gigachat from its api base and drop the dead
gigachat_models elif branch
- constants: register the gigachat api base in openai_compatible_endpoints
- forward unrouted /gigachat/* requests with env credentials like other passthrough providers (the old fallback returned 400 on any request without a routed model, /gigachat/models included)
- fix basedpyright budget breaches across the gigachat provider, common_request_processing, and llm_passthrough_endpoints with real narrowing, no new suppressions
- add regression tests for the fallback target, auth header, and model-less endpoints
* fix(otel/v2): detach credential-routed tenant spans into their own trace
Multi-tenant OTel v2 routes a team or key's LLM-call span to that tenant's
own vendor account (New Relic, Arize, Langfuse, Weave) via dynamic OTLP
credential headers, while the request-root, auth, and db spans stay on the
operator's default backend. The span was still parented into the request
trace, so the tenant account received a child whose parent it never got,
and New Relic rendered it as a fragmented trace with a missing parent.
Detach a credential-routed span the same way a project-routed (Phoenix)
span already detaches: root a fresh trace in the tenant account and link
back to the request trace for correlation. Service-name routing keeps
parenting, since it only relabels service.name on the same operator
backend where the parent is present.
Guard the detach on the callback actually owning an OTLP exporter the
credentials can reach: a callback owning only a console or in_memory
exporter has nowhere to stamp them, so the span would export to the
default backend unchanged and detaching would orphan it on the very
backend that holds its parent. In that case warn once and keep the
default tracer.
* fix(otel/v2): derive tenant-route routability from resolved exporter transport
A denylist classified an owned exporter as routable whenever its kind was
not console/in_memory, so a typo'd or unavailable kind (e.g. "otlp",
"grcp") passed the check while _exporter_from_spec falls it back to a
header-ignoring console exporter. Detaching such a span would root a fresh
trace that only ever reaches the operator console, never the tenant
backend, orphaning it on both sides.
Route on a shared exporter_transport() predicate that resolves the kind the
same way _exporter_from_spec builds it (registered factories + otlp_http
aliases -> http, otlp_grpc aliases -> grpc, else headerless), so an
unresolvable kind is headerless and stays parented. Fixes the same latent
gap in project routability.
The componentized chart's Ingress renders a fixed path set: the UI paths, a
hardcoded gateway prefix list mirroring gateway/routes/allowlist.py, and a `/`
Prefix catch-all to the backend. values.yaml exposes no path knob, so an
operator who needs a route the chart does not know about has to fork the
template.
That prefix list is a snapshot of the data plane at release time and drifts
from it. `/watsonx` is in GATEWAY_PATH_PREFIXES with no ingress rule and no
backend prefix, so `/watsonx/*` falls into the catch-all, reaches the backend,
and 404s. A provider passthrough prefix added upstream after the chart version
an operator runs has the same shape, and a custom
general_settings.pass_through_endpoints route has a path only the operator
knows, so the chart can never ship a rule for it at all.
ingress.extraPaths takes a list of {path, service, pathType} entries rendered
in addition to the built-in paths, never in place of them. They render after
every built-in path and before the backend catch-all, and an entry repeating a
path the chart already routes is rejected: duplicates within one rule are
resolved by position or by controller-specific tie breaking, so an entry at
`/` Prefix would swallow the whole backend management API rather than add to
it. `service` picks the component Service — gateway by default, or backend or
ui — so an operator does not have to reconstruct release-templated Service
names, and `pathType` defaults to Prefix. A non-mapping entry, an entry with
no path, an unknown service, an unknown pathType, and a duplicate each abort
the render naming the offending index rather than emitting an Ingress that
misroutes traffic.
The duplicate check needs the built-in paths as data, so the UI paths move
from six literal YAML blocks into a $uiPaths list rendered by one range, and
their explanatory comments move from the rendered manifest into template
comments. The Ingress object this produces is unchanged: the same 92 paths in
the same order with the same backends, verified by parsing both renders.
The target component still answers only the paths its own route allowlist
keeps, so this makes a route routable, not servable.
The Add Model page's connection test dialog rendered a literal ", ]" beside the Close button, left over from converting an array of footer buttons into JSX children.
The driver retries rate-limit-shaped failures twice with a 65s backoff, but
RATE_LIMIT_SHAPED_RE only matches 429 / rate limit / too many requests /
throttled / CLI timeout. A saturated provider does not always say any of
those: litellm-e2e-pr build 182 turned a green cell red on
status 503: litellm.ServiceUnavailableError: BedrockException -
{"message":"Bedrock is unable to process your request."}
which matched nothing and so was never retried.
Add TRANSIENT_UPSTREAM_SHAPED_RE for 503, 529, "service unavailable",
"overloaded" and Bedrock's "unable to process your request", and have the
retry loop ask is_retryable_shaped (either shape) instead of
is_rate_limit_shaped.
Kept as a second pattern rather than widened into the first on purpose. The
conftest feeds RATE_LIMIT_SHAPED_RE into the rate-limit summary, which the
binary-search helper reads to decide whether to lower a provider's request
rate. A 503 says the provider is out of capacity, not that we are asking too
often, so folding it in there would keep ratcheting our rate down against a
condition our rate never caused.
Retry-loop logging now names which shape fired, so a saturated upstream is
distinguishable from a throttled one in the run output.
Verified against the literal failure text from build 182: is_retryable_shaped
is True while is_rate_limit_shaped stays False.
Note this covers CLI-driven rows only. HTTP-probe rows (tool_search,
count_tokens) have no retry layer at all, so the specific cell that failed in
build 182 is still unprotected -- that is a separate change to http_probe.
#34940 widened the mask-in-place safety guard so a Responses-API
`instructions` field (and a combined messages+input body) skips the
PII masking branch. With `on_flagged: "monitor"` that fell straight
through to "allow", so PII that used to be masked now reaches the
model unredacted.
Monitor means "don't block", not "don't redact". Recover the one shape
whose payload is still fully writable: mask it and write the redacted
instructions back into `data["instructions"]` directly, since
apply_redacted_messages_back has no path for that field and would
otherwise fold the instructions text into `data["input"]`.
The combined messages+input and multimodal shapes stay unmasked - both
are unsafe to write back, not merely unwritable - and now log an error
naming the reason instead of passing silently.
No block/allow decision changes: block and inject_system_message keep
the exact outcomes #34940 shipped.
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
An Admin Viewer and a group whose name contains a slash both reached an
enabled Set budget action that could only ever come back 403 or 404. Gate
the row actions on proxy admin and on the name being addressable, with the
reason in the tooltip.
Model access group budgets shipped API-only, so the only way to give a group a
budget was a curl. Adds an Access Group Budgets tab under Models & Endpoints
that lists every group with the spend drawn against its shared pool, and a
modal to set, edit or clear the budget.
/access_group/list now carries each group's budget and spend inline, so the
table renders from one read instead of one follow-up request per row.
The first pass accepted any empty completion whose finish_reason was "length",
which also swallowed a fallback that produced nothing at all. Require the
response to have billed completion tokens as well, so empty content is
accepted only when the budget was demonstrably spent on non-visible reasoning.
Asserts on completion_tokens rather than reasoning_tokens because the latter
is provider-optional; with empty content and a refusal of null, consumed
completion tokens are reasoning by elimination, since visible text would be
content. Both counts are reported in the failure message.
Folds the three body accessors onto one _parsed helper instead of re-parsing
per call, and adds completion_tokens_of / reasoning_tokens_of alongside.
Multi-window budgets (budget_limits on keys/teams) currently keep window
spend only in cache. Every cold or expired counter recomputes the window
by aggregating LiteLLM_SpendLogs, which has no usable index for that
query and saturates the DB on large tables (#35766).
This adds a LiteLLM_BudgetWindowSpend table holding one row per
configured window, keyed (entity_type, entity_id, window_duration),
with window_start identifying the period the spend belongs to.
Follow-up PRs maintain these rows from the spend update writer and move
window budget enforcement reads onto them.
The LLM classifier's cost was recorded on the routing decision but never
reached any savings surface: per-request autorouter_savings stayed gross
and the session rollup recorded only the served request's spend, so
/auto_router/benchmarks overstated savings and understated routed spend.
Net the classifier cost into the savings figure at its one computation
owner and fold it into the rollup turn's spend, keeping
baseline_spend = spend + saved_spend. The response header's numeric
guard now shares the same reader.
Fixes#38816
max_tokens=64 caps reasoning plus visible output on gpt-5.5, so the fallback
target can legitimately return finish_reason="length" with empty content.
litellm-e2e build 90 hit exactly that: the response cost of $0.002005 backs
out to 64 completion tokens at gpt-5.5's $3e-05/token, i.e. the whole budget
spent reasoning about "say hi" with none left to answer. The fallback itself
worked -- 200, served by gpt-5.5-2026-04-23, x-litellm-attempted-fallbacks
present -- so the only thing that failed was an assertion about OpenAI's token
budgeting rather than about routing.
Raise the reliability helper's budget to 512 and accept empty content only
when finish_reason is "length". Empty content under any other finish_reason
still fails, so the tests keep catching a fallback that returns nothing for a
reason we do control.
The relaxed assertion lives in the helper both reliability fallback tests
share, so test_timeout_routes_to_fallback is covered too; it has the same
shape and had not tripped yet.
