scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-30 01:52:18 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 114
litellm/tests/test_litellm/llms
History
tin-berri ca0b951a43
feat(spend): report prompt caching savings as total and gateway-attributed (#38134) 
* feat(spend): report prompt caching savings as total and gateway-attributed

`prompt_caching_savings_spend` credited every cached request, including caching a
client asked for with its own `cache_control` and caching a provider does implicitly,
so the number overstated what the gateway had any hand in.

Gating that column in place would have fixed the overstatement by changing what the
column means, leaving rows written before the change saying "all caching savings" and
rows after saying "gateway-injected only" with nothing to tell them apart, and forcing
a decision about rewriting history. It also breaks the cache-leakage estimate on the
dashboard, whose numerator would be gated while its denominator, the cached token
counts, would not, so the rate it extrapolates from would be quietly diluted.

Report both instead. `prompt_caching_savings_spend` keeps meaning every net dollar
caching saved, which is what a customer means by "what did caching save me", and the
new `gateway_injected_caching_savings_spend` carries the subset litellm caused by
injecting the breakpoints itself. Both are derived from the same marker, so this
changes what is done with it rather than how it is obtained.

The attributed figure is normally the smaller of the two, being a subset of the same
requests, but not always: a request that writes cache it never reads has negative net
savings, and excluding such a request can lift the attributed figure above the total.

Also stops the marker riding into a fallback leg. The fallback rebuild spread the
failed attempt's metadata forward, so a deployment that injected nothing inherited the
marker and was credited anyway, which silently restored the very overstatement this
separates out.

* fix(bedrock): credit gateway caching where the tool cachePoint is placed (#38478)

The savings marker records breakpoints litellm placed, and a tool_config
injection point becomes one only in the converse transform, and only when the
request carries tools. The prompt hook cannot see either condition, so marking
on the point's presence credited request shapes that cached nothing, while
Bedrock tool caching the gateway did cause went uncredited.

Record it at the placement site instead. The marker's reader also resolves its
bucket by value now: litellm_params declares litellm_metadata as None on every
request, so asking the shared name resolver named a bucket that was not there
and the mark was dropped.
2026-08-28 00:19:06 -07:00
..
a2a/chat fix(a2a): populate response usage in a2a chat transformation (#31980) 2026-07-03 09:28:36 +05:30
aiml/image_generation test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
amazon_nova/chat test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
anthropic fix(anthropic): drop and self-heal empty thinking blocks on /v1/messages (#38625) 2026-08-27 21:41:56 -07:00
apiserpent test: unwind environment writes in tests/test_litellm with monkeypatch (#37806) 2026-08-21 20:28:37 -07:00
azure fix(openai): drop tool_reference parts from tool messages at the chat boundary 2026-08-26 23:43:41 -07:00
azure_ai Merge remote-tracking branch 'origin/litellm_internal_staging' into litellm_fix_model_router_spend_log_model 2026-08-25 10:55:53 -07:00
base_llm fix(exceptions): keep a refused connection an APIConnectionError 2026-08-27 21:00:24 -07:00
baseten/chat style: run black formatter on files from main merge 2026-04-17 13:02:59 -07:00
bedrock feat(spend): report prompt caching savings as total and gateway-attributed (#38134) 2026-08-28 00:19:06 -07:00
bedrock_mantle Merge remote-tracking branch 'origin/litellm_internal_staging' into litellm_registry_audit_bedrock_sol_anthropic_1hr 2026-08-26 19:02:56 +00:00
black_forest_labs test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
bytez/chat test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
cerebras fix(cerebras): add max_retries and extra_headers to get_supported_openai_params (#36601) 2026-08-25 14:10:56 -07:00
chat test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
chatgpt test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
cloudflare fix(cloudflare): route native Workers AI provider through OpenAI-compatible endpoint (#31053) 2026-06-23 12:09:11 -07:00
cohere test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
cometapi/chat test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
compactifai fix(exceptions): map upstream status codes for providers with no exception_type branch 2026-08-26 00:47:28 -07:00
crusoe test: unwind environment writes in tests/test_litellm with monkeypatch (#37806) 2026-08-21 20:28:37 -07:00
custom_httpx fix(transcription): drop words from srt/vtt replies when synthesis falls back 2026-08-27 13:36:31 -07:00
dashscope test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
databricks fix(model_prices): absorb Databricks/Z.AI and xAI registry PRs, add Together and Azure deprecation dates 2026-08-27 13:16:12 +00:00
datarobot test: unwind environment writes in tests/test_litellm with monkeypatch (#37806) 2026-08-21 20:28:37 -07:00
deepgram test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
deepinfra test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
deepseek fix: collapse image_url blocks whose payload lacks a url instead of forwarding them 2026-08-26 13:19:06 -07:00
docker_model_runner test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
elevenlabs test: avoid provider test module name collision 2026-04-29 22:36:57 -07:00
fal_ai test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
fastcrw/search chore(oss): litellm oss staging 150626 (#30463) 2026-06-16 12:06:41 -07:00
featherless_ai/chat test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
fireworks_ai fix(fireworks_ai): stop using the trace id as the session affinity key (#35754) 2026-08-26 18:33:56 -04:00
gdc/chat test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
gemini Merge pull request #38457 from BerriAI/litellm_realtime_audio_output_tokens 2026-08-27 15:23:46 -07:00
github_copilot test(realtime): pin mode-only transcription detection and correct a stale docstring 2026-08-26 12:33:46 -07:00
gradient_ai test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
groq fix: omit thinking.type=disabled for always-on thinking Claude models (#37510) 2026-08-21 10:27:26 -07:00
heroku style: run black formatter on files from main merge 2026-04-17 13:02:59 -07:00
hosted_vllm fix(anthropic): stop signing replayed thinking blocks and strip reasoning_content 2026-08-22 14:27:20 -07:00
huggingface test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
inception test: unwind environment writes in tests/test_litellm with monkeypatch (#37806) 2026-08-21 20:28:37 -07:00
jina_ai/embedding test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
langflow fix(exceptions): map upstream status codes for providers with no exception_type branch 2026-08-26 00:47:28 -07:00
lemonade test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
litellm_proxy test: remove tests that never execute 2026-08-12 10:45:38 -07:00
llamafile/chat style: run black formatter on files from main merge 2026-04-17 13:02:59 -07:00
lm_studio style: run black formatter on files from main merge 2026-04-17 13:02:59 -07:00
manus test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
meta_llama test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
minimax fix(minimax): attach MINIMAX_API_KEY on anthropic messages requests 2026-08-26 12:19:37 -07:00
mistral test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
modelscope test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
moonshot fix(moonshot, together_ai): send the reasoning effort Kimi K3 accepts (#38611) 2026-08-27 20:46:46 -07:00
nebius test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
neosantara Litellm oss staging 050626 (#29774) 2026-06-05 13:51:51 -07:00
nimble/search feat(search): add Nimble as a search provider (#36347) 2026-08-14 17:09:58 -07:00
novita/chat test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
nscale/chat test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
nvidia_nim/rerank test: drop rerank package marker colliding with voyage test package 2026-08-11 23:37:27 -07:00
nvidia_riva test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
oci test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
ocr test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
ollama Merge remote-tracking branch 'origin/litellm_internal_staging' into litellm_ruff_dead_test_code 2026-08-24 09:46:56 -07:00
oobabooga/chat test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
openai fix(openai): drop tool_reference parts from tool messages at the chat boundary 2026-08-26 23:43:41 -07:00
openai_like Merge remote-tracking branch 'origin/litellm_internal_staging' into litellm_scx_ai_provider 2026-08-21 17:27:49 +10:00
openrouter test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
ovhcloud test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
parallel_ai test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
parasail Litellm oss staging 080626 (#29932) 2026-06-08 13:49:52 -07:00
pass_through test: remove four mirror test files that exercise none of their module 2026-07-25 10:57:19 -07:00
perplexity test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
pg_vector/vector_stores test: require a match= on broad pytest.raises, and drop duplicate parametrize cases (#37769) 2026-08-20 20:24:49 -07:00
publicai test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
ragflow/chat test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
recraft test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
reducto Litellm oss staging 04 21 2026 2 (#26569) 2026-05-20 21:25:19 -07:00
runwayml fix(runwayml): skip progress scaling when Runway returns a null progress 2026-08-24 12:20:00 -07:00
s3_vectors Fix extra body error 2026-04-29 08:34:31 +05:30
sagemaker test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
sambanova Feat/sambanova embeddings (#13308) 2025-08-12 17:15:26 -07:00
sap style: run black formatter on files from main merge 2026-04-17 13:02:59 -07:00
scaleway Feature/add audio support for scaleway (#26110) 2026-04-20 14:49:41 -07:00
snowflake chore(oss): litellm oss staging 120626 (#30292) 2026-06-12 09:49:25 -07:00
soniox fix(gemini): ignore non-string response_format when deciding on word timestamps 2026-08-27 13:15:17 -07:00
stability test: require a match= on broad pytest.raises, and drop duplicate parametrize cases (#37769) 2026-08-20 20:24:49 -07:00
tencent test(tencent): drive thinking tests off the real cost map instead of patched internals 2026-08-26 11:34:06 -07:00
tinyfish test: require a match= on broad pytest.raises, and drop duplicate parametrize cases (#37769) 2026-08-20 20:24:49 -07:00
together_ai/chat fix(moonshot, together_ai): send the reasoning effort Kimi K3 accepts (#38611) 2026-08-27 20:46:46 -07:00
valkey/vector_stores feat(vector_stores): add Valkey as a managed vector store provider (#37002) 2026-08-18 21:45:22 +00:00
vercel_ai_gateway test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
vertex_ai Merge branch 'litellm_internal_staging' of https://github.com/BerriAI/litellm into litellm_gemini_maps_grounding_cost 2026-08-26 16:27:40 -07:00
volcengine test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
voyage test: require a match= on broad pytest.raises, and drop duplicate parametrize cases (#37769) 2026-08-20 20:24:49 -07:00
wandb test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
watsonx Merge remote-tracking branch 'origin/litellm_internal_staging' into litellm_ruff_dead_test_code 2026-08-24 09:46:56 -07:00
xai fix(model_prices): absorb Databricks/Z.AI and xAI registry PRs, add Together and Azure deprecation dates 2026-08-27 13:16:12 +00:00
you_com test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
zai test: stop the zai tests from leaking env and litellm globals into the session (#37834) 2026-08-21 22:10:34 -07:00
test_cache_control_and_reasoning.py test: drop the cwd-relative sys.path.insert calls from the test suite (#37802) 2026-08-22 09:25:58 -07:00
test_file_content_block.py Litellm oss staging (#28161) 2026-05-18 16:27:44 -07:00
test_file_search_responses.py fix(tests): use canonical litellm_enterprise import path (#27699) 2026-05-12 12:32:57 -07:00
test_lifecycle_fix.py style: run black formatter on files from main merge 2026-04-17 13:02:59 -07:00
test_polling_url_origin_match.py chore(security): close two unaddressed SSRF cases 2026-05-01 18:43:47 +00:00
test_predibase_transformation.py refactor(predibase): migrate transform_request and transform_response… (#25249) 2026-04-27 08:58:22 +05:30