mirror of
https://github.com/BerriAI/litellm.git
synced 2026-09-12 23:01:41 +00:00
6a0fab4498
The /v1/responses streaming surface had three OpenAI-spec violations that
strict deserializers (Grok Build CLI, OpenAI SDK) reject:
1. sequence_number was dropped from the wire. The completion-transformation
streaming iterator set event.__dict__["sequence_number"] = N after
constructing the Pydantic model, but that mutates the instance dict
without updating __pydantic_extra__, so model_dump() silently omits the
field. 13 mutation sites converted to constructor kwargs; an additional
5 event constructors that never set sequence_number at all (OutputTextDone,
ContentPartDone, OutputTextAnnotationAdded, ReasoningSummaryTextDelta,
ResponseCompleted) now do.
2. response.completed.response.text was hardcoded to {}. The spec requires
text.format to be present. Default to {"format": {"type": "text"}}
when the request didn't supply a text config; honor the request value
otherwise.
3. usage.input_tokens_details and usage.output_tokens_details were dropped
when upstream didn't provide prompt_tokens_details / completion_tokens_details.
The spec requires both fields on every response.completed.usage. Both now
always emit, defaulting to zero-valued details (cached_tokens=0,
reasoning_tokens=0). This unblocks Anthropic chat-completion routes,
which omit completion_tokens_details on tool-only function-call responses.
Verified end-to-end with Grok Build CLI 0.2.2 → LiteLLM → claude-haiku-4-5:
grok -p 'hi' --model litellm-haiku --output-format json returns clean JSON
with exit=0 and zero stderr errors. Confirmed across single-word, arithmetic,
multi-line, and 'grok models' catalog probes.
Tests:
- 4 contract-pinning usage tests updated to assert the new always-emit shape.
- 1 streaming-tool-call test that asserted the buggy __dict__ pattern updated
to read sequence_number from model_dump() (the actual wire payload).
- 1 new regression test pins the Grok contract: every streaming event must
carry sequence_number in model_dump() output, monotonic and unique.
|
||
|---|---|---|
| .. | ||
| agent_tests | ||
| audio_tests | ||
| basic_proxy_startup_tests | ||
| batches_tests | ||
| benchmarks | ||
| code_coverage_tests | ||
| documentation_tests | ||
| enterprise | ||
| guardrails_tests | ||
| image_gen_tests | ||
| integration | ||
| litellm | ||
| litellm-proxy-extras | ||
| litellm_core_utils | ||
| litellm_utils_tests | ||
| llm_responses_api_testing | ||
| llm_translation | ||
| load_tests | ||
| local_testing | ||
| logging_callback_tests | ||
| mcp_tests | ||
| multi_instance_e2e_tests | ||
| ocr_tests | ||
| old_proxy_tests/tests | ||
| openai_endpoints_tests | ||
| otel_tests | ||
| pass_through_tests | ||
| pass_through_unit_tests | ||
| proxy_admin_ui_tests | ||
| proxy_behavior | ||
| proxy_e2e_anthropic_messages_tests | ||
| proxy_security_tests | ||
| proxy_unit_tests | ||
| router_unit_tests | ||
| scim_tests | ||
| search_tests | ||
| spend_tracking_tests | ||
| store_model_in_db_tests | ||
| test_litellm | ||
| unified_google_tests | ||
| vector_store_tests | ||
| windows_tests | ||
| __init__.py | ||
| _flush_vcr_cache.py | ||
| _vcr_conftest_common.py | ||
| _vcr_redis_persister.py | ||
| eval_swe_bench.py | ||
| gettysburg.wav | ||
| large_text.py | ||
| openai_batch_completions.jsonl | ||
| README.MD | ||
| test_budget_management.py | ||
| test_callbacks_on_proxy.py | ||
| test_config.py | ||
| test_debug_warning.py | ||
| test_default_encoding_non_root.py | ||
| test_end_users.py | ||
| test_entrypoint.py | ||
| test_fallbacks.py | ||
| test_gpt5_azure_temperature_support.py | ||
| test_health.py | ||
| test_keys.py | ||
| test_litellm_proxy_responses_config.py | ||
| test_logging.conf | ||
| test_models.py | ||
| test_new_vector_store_endpoints.py | ||
| test_openai_endpoints.py | ||
| test_organizations.py | ||
| test_otel_thread_leak.py | ||
| test_passthrough_endpoints.py | ||
| test_presidio_latency.py | ||
| test_proxy_server_non_root.py | ||
| test_ratelimit.py | ||
| test_resource_cleanup.py | ||
| test_service_logger_otel.py | ||
| test_spend_logs.py | ||
| test_team.py | ||
| test_team_logging.py | ||
| test_team_members.py | ||
| test_users.py | ||
In total litellm runs 1000+ tests
[02/20/2025] Update:
To make it easier to contribute and map what behavior is tested,
we've started mapping the litellm directory in tests/test_litellm
This folder can only run mock tests.