mirror of
https://github.com/BerriAI/litellm.git
synced 2026-09-16 23:41:43 +00:00
d4efaf8cf9
`_set_usage_outputs` emits only total / completion / prompt / reasoning
tokens, leaving `LLM_TOKEN_COUNT_PROMPT_DETAILS_CACHE_READ` and
`LLM_TOKEN_COUNT_PROMPT_DETAILS_CACHE_WRITE` defined-but-unused. As a
result, observability backends (Langfuse, Arize Phoenix) cannot display
the prompt-cache breakdown nor apply correct cost calculation for
Anthropic / Bedrock prompt caching (cache reads at 0.1x, cache writes
at 1.25x of standard input).
Read from `usage.prompt_tokens_details` so the change is provider-agnostic.
LiteLLM's `Usage.__init__` already normalizes provider-specific cache
fields onto this object:
- Anthropic `cache_read_input_tokens` -> prompt_tokens_details.cached_tokens
- DeepSeek `prompt_cache_hit_tokens` -> prompt_tokens_details.cached_tokens
- OpenAI native cached_tokens -> prompt_tokens_details.cached_tokens
- Anthropic `cache_creation_input_tokens` -> prompt_tokens_details.cache_creation_tokens
Mapping:
- prompt_tokens_details.cached_tokens
-> llm.token_count.prompt_details.cache_read
- prompt_tokens_details.cache_creation_tokens
-> llm.token_count.prompt_details.cache_write
Note: PR #24112 introduced a similar `cached_tokens` block in
litellm_oss_staging_03_21_2026 but has not yet landed on main; PR #26506
on litellm-oss-staging-04-25-2026 refactored the function without
including it. This change brings the emission directly to main with
extended cache_creation coverage.
Tests:
- Anthropic: both cache_read and cache_write emitted
- OpenAI: only cache_read emitted (no cache_write concept)
- DeepSeek: prompt_cache_hit_tokens normalized -> cache_read emitted
- No-cache: neither attribute emitted
|
||
|---|---|---|
| .. | ||
| agent_tests | ||
| audio_tests | ||
| basic_proxy_startup_tests | ||
| batches_tests | ||
| benchmarks | ||
| code_coverage_tests | ||
| documentation_tests | ||
| enterprise | ||
| guardrails_tests | ||
| image_gen_tests | ||
| 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_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 | ||
| 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.