Two independent fixes for pre-existing test failures on main:
1. Anthropic streaming: The sync __next__ method used a simple
holding_chunk pattern that lost chunks when multiple events needed
to be returned. Refactored to use the same chunk_queue approach as
the async __anext__ method. Also fixed tests that used ModelResponse
(which defaults finish_reason to 'stop') instead of ModelResponseStream.
2. Azure GPT-5.1 logprobs: The base OpenAI class includes logprobs for
gpt-5.1+ models, but Azure hasn't verified support for gpt-5.1.
Added explicit removal of logprobs/top_logprobs for gpt-5.1 (non-5.2)
models in the Azure config.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace **kwargs with explicit tools and system parameters to match
the BaseTokenCounter.count_tokens abstract method signature.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- OpenAICredentials.api_base is always non-None due to fallback default,
so type it as str instead of Optional[str]
- Remove unreachable elif vertex_ai block in create_file(); vertex_ai is
already handled by ProviderConfigManager via provider_config path
Suppress noisy error log fired every cron tick when spend log cleanup
is simply not configured. _should_delete_spend_logs already logs the
specific reason at the right level (info for None, warning for
invalid value), so the redundant blanket error log in
cleanup_old_spend_logs is removed.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Verifies that vertex_ai gemini models route to
aiplatform.googleapis.com instead of
generativelanguage.googleapis.com, preventing
regressions if the branch ordering changes.
- Only release distributed lock in finally if it was actually acquired;
prevents spurious Redis release_lock calls on early returns
- Treat bare integer maximum_spend_logs_retention_period as days (e.g. 3 → "3d")
instead of silently failing with a ValueError
- Elevate "Skipping cleanup" log from info to error so misconfigured
retention settings are visible without verbose logging
- Add tests for all three fixes
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Move get_openai_credentials() to litellm/llms/openai/common_utils.py
and get_azure_credentials() to litellm/llms/azure/common_utils.py
so they can be reused by batches/main.py and other modules.
Signatures now take individual params instead of GenericLiteLLMParams.
Extract repeated OpenAI and Azure credential resolution logic into
_get_openai_credentials() and _get_azure_credentials() helper functions,
reducing ~270 lines of duplicated code across 5 file operations.
Also removes dead Vertex AI code path in create_file that was unreachable
since ProviderConfigManager.get_provider_files_config() handles it first.
Add OpenRouter image edit docs to both the provider page and the
main image_edits reference page, including supported models, parameter
mappings (size→aspect_ratio, quality→image_size), usage examples,
proxy configuration, and a note about 4K quality model support.
- Add created_at field to MCPServer type (was missing)
- Map created_at from LiteLLM_MCPServerTable in build_mcp_server_from_table()
- Use server.created_at and server.updated_at instead of datetime.now() in _build_mcp_server_table() and health check table builder
- Add regression tests to verify timestamps are preserved through round-trip conversions
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The parameter was declared in completion() signature but not passed
to acompletion_with_mcp, causing per-request JSON schema validation
to silently fall back to the global default when MCP tools are present.
Replaces the skip_route_check approach from PR #22662 with a configurable
opt-in flag. By default, common_checks() is not run for custom auth flows,
preserving backwards compatibility with pre-#22164 behavior.
Users who want budget/team/route enforcement on custom auth can enable it:
general_settings:
custom_auth_run_common_checks: true
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
