scriptoriumadmin
  • Joined on 2026-03-01
scriptoriumadmin synced commits to refs/pull/47223/merge at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:27 +00:00
da80552897 test(translation): add sonnet 5.5 converse and gemini direct cross-model mid rows (#47328)
c4d24ec2f1 test(integration): remove BUG skips for 3 tests that pass on main (#47364)
daa133da5c fix(mcp)!: bill both model calls of a chat completions MCP auto-execute on one spend row (#47422)
7b82b2a3c3 feat(bedrock): route token counts from model metadata (#47441)
Compare 96 commits »
scriptoriumadmin synced commits to refs/pull/47225/head at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:27 +00:00
5b8889142d fix(otel v2): stamp the stream the caller got on the trace root
4999477148 fix(otel v2): keep redacted messages off the Arize AX and Phoenix trace root
Compare 2 commits »
scriptoriumadmin synced commits to refs/pull/47231/merge at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:27 +00:00
90af729933 fix(cost-map): enable function calling + tool choice for fireworks deepseek-v4p1-flash-1017 (#47326)
c82e8244a5 fix(count_tokens)!: count tools the way DeepSeek V4 and Qwen3.5 chat templates render them (#47173)
f8f6a01cae feat(ui): show kubernetes pods on the model page (#44666)
f85926a652 feat(router): keep a session on one kubernetes pod (#44393)
Compare 27 commits »
scriptoriumadmin synced commits to refs/pull/47181/merge at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:26 +00:00
7f9cf5d182 chore: merge main into litellm_access_group_user_names
d29200fa05 fix(vertex_ai): bill per-character text embeddings on Google's billableCharacterCount (#47460)
8b9cd12a89 test: catch regressions in the validation lines #44798 added (#47420)
3fa1587f43 fix(logging): do not stamp a provider handoff when pre_call runs for a proxy-only failure (#47456)
Compare 72 commits »
scriptoriumadmin synced commits to refs/pull/47182/head at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:26 +00:00
15544283f6 fix(logging): keep the catalog model_map_key, report only the deployment's own rates
scriptoriumadmin synced commits to refs/pull/47189/head at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:26 +00:00
9888723066 fix(proxy): scope the ui-session spread disables to the marker binding (basedpyright ceiling)
d07c2bb46b fix(proxy): make the ui-session marker parameter object-typed and scope the db-metadata disables (basedpyright ceiling)
f0e3bdeb42 fix(proxy): keep the metadata clear-to-none guard type-true and scope disables to the untyped rows (basedpyright ceiling)
fc1d63ab2a test: assert outside the raises block for the temp-budget owner guard (ruff PT012)
a76323cb7c fix(proxy): tolerate an explicit metadata=None in key-update metadata normalization
Compare 19 commits »
scriptoriumadmin synced commits to refs/pull/47189/merge at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:26 +00:00
9888723066 fix(proxy): scope the ui-session spread disables to the marker binding (basedpyright ceiling)
d07c2bb46b fix(proxy): make the ui-session marker parameter object-typed and scope the db-metadata disables (basedpyright ceiling)
f0e3bdeb42 fix(proxy): keep the metadata clear-to-none guard type-true and scope disables to the untyped rows (basedpyright ceiling)
fc1d63ab2a test: assert outside the raises block for the temp-budget owner guard (ruff PT012)
Compare 15 commits »
scriptoriumadmin synced commits to refs/pull/47190/head at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:26 +00:00
9995e2fd99 chore: merge main into batch reconciliation fix
ae74c53707 fix(batch): preserve recoverable jobs on unrelated resource errors
0e7c4908bb fix(proxy): price the estimated input tokens on a client-cancelled (499) request's spend row (#47436)
1660c8ba1f feat(key_management)!: apply default_key_generate_params.allowed_routes on key generation (#41236)
c6c303a618 fix(batches): apply the Vertex regional endpoint uplift to batch cost (#47466)
Compare 105 commits »
scriptoriumadmin synced commits to refs/pull/47190/merge at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:26 +00:00
9995e2fd99 chore: merge main into batch reconciliation fix
ae74c53707 fix(batch): preserve recoverable jobs on unrelated resource errors
0e7c4908bb fix(proxy): price the estimated input tokens on a client-cancelled (499) request's spend row (#47436)
1660c8ba1f feat(key_management)!: apply default_key_generate_params.allowed_routes on key generation (#41236)
Compare 92 commits »
scriptoriumadmin synced commits to refs/pull/47193/head at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:26 +00:00
c4a0740885 test: assert the BYOK move path returns clean in both gate files (test-quality TQ001)
d3cd114082 test: mark the BYOK flows async and drop re-import shadowing (ruff F811/PT017)
45ea0fed6a test: pin the interdiction exception types and move capture out of the except block (ruff PT011/PT017)
63b717056e style: ruff-format the api_base repoint predicate (test-quality)
88fb59746e style: ruff-format the model-management tests
Compare 18 commits »
scriptoriumadmin synced commits to refs/pull/47193/merge at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:26 +00:00
2e427c6247 chore(deps): require anyio>=4.14.0 to rule out the lock-waiter deadlock (#47378)
f07596a5fd fix(helm): stop litellm-helm from injecting placeholder models into every install (#47056)
d7620c77a1 fix(openai): drop unsupported reasoning effort for gpt-6.1-sol (#43935)
40e4ade6a6 feat(proxy): opt-in self-serve budget edits for key owners on /key/update (#44763)
Compare 40 commits »
scriptoriumadmin synced commits to refs/pull/47178/merge at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:26 +00:00
7c3d5a86ef fix(proxy): price a billed failure with Logging.get_router_model_id and the deployment's base_model like the success path
b67003162a test(integration): wait for every worker to report the websocket session cap instead of sleeping (#47412)
1525e4bce5 feat(cli): keep the claude.ai login with lite claude --subscription and save pkce in config (#47385)
09c6ee3ac4 Merge remote-tracking branch 'origin/main' into litellm_lit145_timeout_input_cost
Compare 66 commits »
scriptoriumadmin synced commits to refs/pull/47181/head at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:26 +00:00
7f9cf5d182 chore: merge main into litellm_access_group_user_names
d29200fa05 fix(vertex_ai): bill per-character text embeddings on Google's billableCharacterCount (#47460)
8b9cd12a89 test: catch regressions in the validation lines #44798 added (#47420)
3fa1587f43 fix(logging): do not stamp a provider handoff when pre_call runs for a proxy-only failure (#47456)
f98168e73d test(e2e/perf): read the memory baseline after a one minute warm-up (#47455)
Compare 103 commits »
scriptoriumadmin synced commits to refs/pull/47188/head at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:26 +00:00
22f8a0b206 fix(proxy): move the iterative path decode into a helper with honest receipts
291b4e5296 fix(proxy): make the iterative path decode loop-carried (basedpyright reportGeneralTypeIssues)
4a1c7b7053 fix(proxy): authorize the openai passthrough on the decoded forward path
c02bcce7ad test: assert observables in the openai passthrough gate probes (TQ001)
7edc221928 style: ruff-format the passthrough gate files
Compare 19 commits »
scriptoriumadmin synced commits to refs/pull/47183/head at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:26 +00:00
429093d6f3 test(decisions): assert the exact BadRequestError for the rejected-questions case
2ebf82b36d fix(decisions): log the caller request at the SDK entry so requests rejected before prepare_call are recorded too
Compare 2 commits »
scriptoriumadmin synced commits to refs/pull/47187/merge at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:26 +00:00
b690225956 fix(proxy): resync configuration after Redis reconnect (#43313)
2d181e6633 fix(token_counter): count DeepSeek V4 text with DeepSeek's own tokenizer (#47353)
d56ff281de fix(router): key prompt caching affinity on tools and the /v1/messages system prompt (#46411)
fb555b00dc fix(proxy): price a billed failure row at the deployment's own rate like the success path (#47437)
Compare 87 commits »
scriptoriumadmin synced commits to refs/pull/47188/merge at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:26 +00:00
2e427c6247 chore(deps): require anyio>=4.14.0 to rule out the lock-waiter deadlock (#47378)
f07596a5fd fix(helm): stop litellm-helm from injecting placeholder models into every install (#47056)
d7620c77a1 fix(openai): drop unsupported reasoning effort for gpt-6.1-sol (#43935)
40e4ade6a6 feat(proxy): opt-in self-serve budget edits for key owners on /key/update (#44763)
Compare 46 commits »
scriptoriumadmin synced commits to refs/pull/47191/merge at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:26 +00:00
da80552897 test(translation): add sonnet 5.5 converse and gemini direct cross-model mid rows (#47328)
c4d24ec2f1 test(integration): remove BUG skips for 3 tests that pass on main (#47364)
daa133da5c fix(mcp)!: bill both model calls of a chat completions MCP auto-execute on one spend row (#47422)
7b82b2a3c3 feat(bedrock): route token counts from model metadata (#47441)
Compare 93 commits »
scriptoriumadmin synced commits to refs/pull/47151/merge at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:25 +00:00
783cf06f31 fix(sdk): decide MCP gateway routing without importing the proxy handler (#45961)
6c4c134f23 fix(secret_managers)!: resolve oidc/file and oidc/env refs in admin-configured api_key per request (#47341)
923b7d1784 test(integration): read router cooldown failure rows from the proxy that served them (#47457)
ca329a7f13 chore(bedrock): add Model EOL deprecation dates for Palmyra X4, X5 and Pegasus v1.2 (#47462)
Compare 77 commits »
scriptoriumadmin synced commits to refs/pull/47158/merge at scriptoriumadmin/litellm from mirror 2026-10-20 05:06:25 +00:00
8b9cd12a89 test: catch regressions in the validation lines #44798 added (#47420)
3fa1587f43 fix(logging): do not stamp a provider handoff when pre_call runs for a proxy-only failure (#47456)
f98168e73d test(e2e/perf): read the memory baseline after a one minute warm-up (#47455)
b2eda10bef perf(router): resolve deployment usage limits once per deployment instead of every request (#47393)
Compare 70 commits »