yucheng-berri
4301ac4940
Merge pull request #41986 from BerriAI/litellm_revert_post_call_guardrail_context
...
revert(guardrails): drop the scoped request conversation and tools from post-call scans (#41220 )
2026-09-19 12:06:30 -07:00
joshua-berri
eda1faba14
Merge pull request #41718 from BerriAI/litellm_mcp_dependency_gate_7737
...
refactor(mcp): upgrade SDK2 while preserving legacy gateway behavior
2026-09-19 18:53:19 +00:00
kerry-berri
c3fa53be8d
Merge pull request #41993 from BerriAI/litellm_remove_auto_merge_price_sync
...
ci: remove auto-merge-price-sync workflow, the Devin sync automation merges price PRs
2026-09-19 11:44:36 -07:00
yujonglee
f081e447a4
Merge pull request #41992 from BerriAI/litellm_deepseek_ocr_defaults
...
fix(ocr): set DeepSeek OCR sampling defaults
2026-09-19 11:42:49 -07:00
kerry-berri
b89a6fba4c
Merge pull request #41833 from BerriAI/litellm-providers/price-sync-openrouter
...
chore(prices): sync OpenRouter prices: 172 models, 2 new
2026-09-19 11:42:34 -07:00
kerry
987af6c66c
ci: remove auto-merge-price-sync workflow, the Devin sync automation merges price PRs
...
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 18:33:06 +00:00
ryan-crabbe-berri
659cef0f57
Merge pull request #40729 from BerriAI/litellm_lit7572_top_keys_user_column
...
fix(ui): show user attribution in Top Virtual Keys usage tables
2026-09-19 11:32:35 -07:00
Yujong Lee
8c21a988b7
fix(ocr): set DeepSeek OCR sampling defaults
...
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 18:31:49 +00:00
berriai-litellm-provider-info-sync[bot]
a04ba30f7d
chore(prices): sync OpenRouter prices: 172 models, 2 new
...
openrouter/~anthropic/claude-fable-latest: supports_web_search
openrouter/~anthropic/claude-haiku-latest: supports_web_search
openrouter/~anthropic/claude-opus-latest: supports_web_search
openrouter/~anthropic/claude-sonnet-latest: supports_web_search
openrouter/~deepseek/deepseek-flash-latest: max_tokens, max_output_tokens, input_cost_per_token, output_cost_per_token, cache_read_input_token_cost
openrouter/~deepseek/deepseek-v4-flash-latest: max_tokens, max_output_tokens, input_cost_per_token, output_cost_per_token, cache_read_input_token_cost
openrouter/~google/gemini-flash-latest: supports_web_search
openrouter/~google/gemini-pro-latest: supports_web_search
openrouter/~moonshotai/kimi-latest: input_cost_per_token, output_cost_per_token, cache_read_input_token_cost
openrouter/~openai/gpt-astra-latest: supports_web_search
openrouter/~openai/gpt-luna-latest: supports_web_search
openrouter/~openai/gpt-mini-latest: supports_web_search
openrouter/~openai/gpt-sol-latest: supports_web_search
openrouter/~openai/gpt-terra-latest: supports_web_search
openrouter/~x-ai/grok-latest: supports_web_search
openrouter/~z-ai/glm-latest: max_tokens, max_output_tokens, input_cost_per_token, output_cost_per_token, cache_read_input_token_cost
openrouter/anthropic/claude-3-haiku: supports_web_search
openrouter/anthropic/claude-fable-5: supports_web_search
openrouter/anthropic/claude-fable-5:batch: supports_web_search
openrouter/anthropic/claude-fable-5.1: supports_web_search
openrouter/anthropic/claude-fable-5.1:batch: supports_web_search
openrouter/anthropic/claude-haiku-4.5: supports_web_search
openrouter/anthropic/claude-haiku-4.5:batch: supports_web_search
openrouter/anthropic/claude-opus-4: supports_web_search
openrouter/anthropic/claude-opus-4.1: supports_web_search
openrouter/anthropic/claude-opus-4.1:batch: supports_web_search
openrouter/anthropic/claude-opus-4.5: supports_web_search
openrouter/anthropic/claude-opus-4.5:batch: supports_web_search
openrouter/anthropic/claude-opus-4.6: supports_web_search
openrouter/anthropic/claude-opus-4.6:batch: supports_web_search
openrouter/anthropic/claude-opus-4.7: supports_web_search
openrouter/anthropic/claude-opus-4.7:batch: supports_web_search
openrouter/anthropic/claude-opus-4.8: supports_web_search
openrouter/anthropic/claude-opus-4.8:batch: supports_web_search
openrouter/anthropic/claude-opus-5: supports_web_search
openrouter/anthropic/claude-opus-5:batch: supports_web_search
openrouter/anthropic/claude-sonnet-4: supports_web_search
openrouter/anthropic/claude-sonnet-4.5: supports_web_search
openrouter/anthropic/claude-sonnet-4.5:batch: supports_web_search
openrouter/anthropic/claude-sonnet-4.6: supports_web_search
openrouter/anthropic/claude-sonnet-4.6:batch: supports_web_search
openrouter/anthropic/claude-sonnet-5: supports_web_search
openrouter/anthropic/claude-sonnet-5:batch: supports_web_search
openrouter/deepseek/deepseek-v4-flash: input_cost_per_token, output_cost_per_token, cache_read_input_token_cost
openrouter/deepseek/deepseek-v4-flash-0731: input_cost_per_token, output_cost_per_token, cache_read_input_token_cost
openrouter/deepseek/deepseek-v4-flash-vision-exp: max_tokens, max_output_tokens
openrouter/deepseek/deepseek-v4-pro: max_tokens, max_output_tokens, input_cost_per_token, output_cost_per_token, cache_read_input_token_cost
openrouter/deepseek/deepseek-v4.1-flash: off_peak_pricing, input_cost_per_token, output_cost_per_token, cache_read_input_token_cost
openrouter/google/gemini-2.5-flash: supports_web_search
openrouter/google/gemini-2.5-flash-image: supports_web_search
openrouter/google/gemini-2.5-flash-lite: supports_web_search
openrouter/google/gemini-2.5-flash-lite:batch: supports_web_search
openrouter/google/gemini-2.5-flash:batch: supports_web_search
openrouter/google/gemini-2.5-pro: supports_web_search
openrouter/google/gemini-2.5-pro-preview: supports_web_search
openrouter/google/gemini-2.5-pro:batch: supports_web_search
openrouter/google/gemini-3-flash-preview: supports_web_search
openrouter/google/gemini-3-flash-preview:batch: supports_web_search
openrouter/google/gemini-3-pro-image: supports_web_search
openrouter/google/gemini-3-pro-image-preview: supports_web_search
2026-09-19 18:31:02 +00:00
ryan-crabbe-berri
82fd632153
test(ui): share one chart bar lookup across Top Virtual Keys tests
...
The key_exists chart test added a second direct DOM lookup for the Recharts bars, which exposes no role or label, and pushed testing-library/no-node-access over its budget (709 > 707). Both chart tests now go through one helper
2026-09-19 11:22:01 -07:00
kerry-berri
fc5f6c4b50
Merge pull request #41964 from BerriAI/litellm_registry_audit_20260919
...
fix(model_prices): drop anthropic deprecation floors and correct azure gpt-4.1-nano retirement date
2026-09-19 11:18:32 -07:00
yucheng-berri
2542ea0320
Merge pull request #41787 from BerriAI/litellm_llm_guard_proxy_call_types
...
fix(llmguard): accept proxy async call types
2026-09-19 11:17:21 -07:00
ryan-crabbe-berri
89bf870225
fix(ui): stop Top Virtual Keys from opening keys that are not in the database
...
/user/daily/activity now reports key_exists on each api key's metadata, true
only when the key is in the active key table that /key/info reads. Top Virtual
Keys renders the Key ID as plain text with an explanatory tooltip and ignores
chart bar clicks when key_exists is false, so deleted keys and CLI/SSO session
keys no longer dead-end on a "Key not found in database" toast
2026-09-19 11:08:21 -07:00
Devin AI
107ec2706b
style(ui): format Top Virtual Keys aggregation test
...
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 11:04:46 -07:00
Devin AI
88799f6f80
fix(ui): fall back to user id in Top Virtual Keys user column
...
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 11:04:46 -07:00
Devin AI
e93fe60512
fix(ui): hide Top Virtual Keys user column when rows carry no user
...
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 11:04:26 -07:00
Devin AI
3d805e5166
fix(ui): show user attribution in Top Virtual Keys usage tables
...
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 11:04:26 -07:00
yucheng-berri
2886b8ee27
Merge pull request #41939 from BerriAI/litellm_responses_missing_input_400
...
* fix(proxy): return 400 instead of 500 for /v1/responses without input
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix(proxy): validate input before starting background responses polling
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix(proxy): validate responses input after prompt template expansion
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
---------
Co-authored-by: yucheng <yucheng@berri.ai>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 11:02:30 -07:00
yucheng
c38dda2b2f
fix(llmguard): drop call types the proxy never routes through moderation
...
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 17:57:47 +00:00
tin-berri
77a4624f13
Merge pull request #41920 from BerriAI/litellm_claude_auto_cache_providers
...
fix(caching): scope automatic breakpoints to supported Claude transports
2026-09-19 12:54:30 -05:00
yucheng-berri
b6758159dd
Merge pull request #41982 from BerriAI/litellm_otel_v2_langfuse_embedding_output
...
* fix(otel v2): summarize embedding vectors as Langfuse observation output
The v2 LLM span built its output only from response choices, so /v1/embeddings
rendered a Langfuse generation with input, usage and cost but a blank output.
Embedding calls now carry an EmbeddingOutput(count, dimensions) summary that the
Langfuse mapper serializes as the observation output, and they are exported with
the embedding observation type instead of generation. Chat and Responses output
mapping is unchanged.
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix(otel v2): keep embedding observations typed as generation in Langfuse
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
---------
Co-authored-by: yucheng <yucheng@berri.ai>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 10:53:13 -07:00
Joshua Valluru
78e1103bb8
fix(ci): preserve shared runner setup time allowance
2026-09-19 10:46:38 -07:00
yucheng
537cdaf487
Revert "Merge pull request #41220 from BerriAI/litellm_post_call_guardrail_context"
...
This reverts commit e40b90bbfad8d5437f55
2026-09-19 17:45:15 +00:00
Joshua Valluru
8f3562ed9c
ci(mcp): consolidate integration tests into shared workflow
2026-09-19 10:41:36 -07:00
yucheng
c4d6c3046e
fix(otel v2): keep embedding observations typed as generation in Langfuse
...
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 17:38:20 +00:00
yuneng-jiang
d0f60fcdec
Merge pull request #41985 from BerriAI/litellm_config_shadows_db_warning
...
feat(proxy): say when a stored setting is ignored because the config file owns it
2026-09-19 10:36:01 -07:00
Yassin Kortam
4487a9915b
Merge pull request #41915 from BerriAI/litellm_distributed_tpm_shared_redis_check
...
fix(router): enforce model tpm limits against shared redis usage across replicas
2026-09-19 10:22:25 -07:00
Yuneng Jiang
7353b779c2
feat(proxy): say when a stored setting is ignored because the config file owns it
...
The config file winning over the database was silent. An admin who had set
a value through the UI and later pinned the same key in the file saw their
stored value quietly stop applying, with nothing said at boot and nothing
said when a later write was refused.
Startup now warns once per key whose stored value differs from the file's,
naming the key and what to do about it. The refusal raised on a write to a
config-owned key carries the same sentence, so the log and the 400 read
identically, and both call out that a stored value exists and will never be
applied. The /config/update refusal gained the same detail.
Keys the file does not declare are untouched: the database still owns them,
and a stored value equal to the file's is not worth a warning.
2026-09-19 10:20:31 -07:00
yujonglee
bd82d73ca1
Merge pull request #41981 from BerriAI/litellm_rust_typed_pyo3
...
refactor(rust): use typed pyo3 APIs instead of getattr/import strings
2026-09-19 10:09:34 -07:00
Tin Chi Lo
2b086dc7aa
fix(caching): scope automatic breakpoints to supported Claude transports
2026-09-19 09:55:38 -07:00
yujonglee
209a780992
Merge pull request #41977 from BerriAI/litellm_rust_sealed_request_textract
...
feat(rust): add Amazon Textract to litellm.ocr and sign provider requests after host hooks
2026-09-19 09:53:07 -07:00
yucheng
c5181f6178
fix(otel v2): summarize embedding vectors as Langfuse observation output
...
The v2 LLM span built its output only from response choices, so /v1/embeddings
rendered a Langfuse generation with input, usage and cost but a blank output.
Embedding calls now carry an EmbeddingOutput(count, dimensions) summary that the
Langfuse mapper serializes as the observation output, and they are exported with
the embedding observation type instead of generation. Chat and Responses output
mapping is unchanged.
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 16:52:25 +00:00
Yujong Lee
619a19b8a2
refactor(rust): use typed pyo3 APIs instead of getattr/import strings
...
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 16:46:34 +00:00
Joshua Valluru
fb56a14cd4
chore(mcp): merge main with unit test timeout safeguards
2026-09-19 09:42:08 -07:00
kerry-berri
5d28684016
Merge pull request #41966 from BerriAI/litellm-providers/price-sync-azure
...
chore(prices): sync Azure prices: 5 models, 5 deprecated
2026-09-19 09:41:53 -07:00
Devin AI
2f1c8669ec
fix(model_prices): drop anthropic "not sooner than" floors from deprecation_date
...
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 16:40:05 +00:00
kerry-berri
6718fd67fb
Merge pull request #41328 from BerriAI/litellm_e2e_cost_calculation_scripted_provider
...
test(integration): literal request/response cost tracking suite on a test-owned cost map
2026-09-19 09:40:04 -07:00
Mateo Wang
f9d4979d21
Merge pull request #41946 from BerriAI/litellm_deflake_20260919
2026-09-19 09:38:25 -07:00
yucheng
13b05af06e
fix(proxy): validate responses input after prompt template expansion
...
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 16:34:53 +00:00
Yujong Lee
eb502824f0
fix stuff
2026-09-19 09:32:46 -07:00
berriai-litellm-provider-info-sync[bot]
196a631835
chore(prices): sync Azure prices: 5 models, 5 deprecated
...
azure/eu/gpt-4.1-nano: deprecation_date
azure/gpt-4.1-nano: deprecation_date
azure/gpt-4.1-nano-2025-04-14: deprecation_date
azure/us/gpt-4.1-nano: deprecation_date
azure/us/gpt-4.1-nano-2025-04-14: deprecation_date
2026-09-19 16:30:59 +00:00
yucheng
012d82d85d
fix(llmguard): scan input and prompt even when messages is present
...
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 16:30:11 +00:00
joshua-berri
f08b787685
Merge pull request #41975 from BerriAI/litellm_fix_azure_snapshot_fallback_test
...
test(utils): isolate dated model fallback from pricing additions
2026-09-19 16:20:51 +00:00
Yujong Lee
c404bed9f0
feat(rust): add Amazon Textract to litellm.ocr and sign provider requests after host hooks
...
Add an aws_textract OCR provider on the Rust route, with no Python path. The
detect-document-text model returns plain lines and analyze-document renders
layout and tables as markdown. Both use Textract's synchronous API, so a
multi-page PDF or TIFF is rejected with an error that names the single-page
limit. A call with no region fails instead of falling back to Bedrock's default
SigV4 covers the request body, and host hooks can rewrite that body before it
is sent. litellm-http now has OutboundRequest, which serializes the body once,
shows those bytes to a RequestSigner and is the only thing a route can send.
Chat, audio transcription and OCR build it after their hooks ran, so a callback
that redacts the body still produces a valid Bedrock or Textract signature
ChatCompletionsAuth and AudioTranscriptionAuth are replaced by
litellm_auth::RequestAuth, and one helper in core turns it into a signed or
unsigned request. Audio transcription now signs only the AWS header set and
rejects a forwarded header that SigV4 computes, the same as chat
The OCR catalog routes aws_textract as Rust required, and the dispatch context
reads the provider from the model prefix so a provider scoped rule can match
2026-09-19 09:11:52 -07:00
Devin AI
e94c9eab03
Merge branch 'main' into litellm_registry_audit_20260919
...
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 16:06:41 +00:00
Devin AI
8f613511cf
test(utils): use a synthetic snapshot date in the dated-to-undated fallback test
...
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 16:03:42 +00:00
yujonglee
362be56bb0
Merge pull request #41969 from BerriAI/litellm_rust_settings_layers
...
refactor(rust): centralize layered settings resolution
2026-09-19 09:00:19 -07:00
yuneng-jiang
6ed9deb4d7
Merge pull request #41931 from BerriAI/litellm_settings_store_immutable
...
fix(proxy): refuse runtime writes to config-owned settings
2026-09-19 09:00:13 -07:00
Yujong Lee
0d0c63dde1
fix(rust): suppress private settings resolver lint
...
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-19 15:48:07 +00:00
Mateo Wang
638e085dba
Merge pull request #41971 from BerriAI/litellm_fix_dated_snapshot_test_exact_key
...
test(cost): point dated snapshot tests at a date the cost map cannot carry
2026-09-19 08:39:20 -07:00
