scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-07 08:26:10 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 36

fix(bedrock): correct max_output_tokens for moonshotai.kimi-k2.5 (#23334)

Browse source 
* fix(bedrock): correct max_output_tokens for moonshotai.kimi-k2.5 from 262144 to 65536

Fixes #22478. The max_output_tokens for all Bedrock moonshotai.kimi-k2.5
variants was incorrectly set to 262144 (the full context window size).
Bedrock enforces input_tokens + max_tokens <= context_window, so sending
max_tokens=262144 causes requests to fail. Corrected to 65536 (64K) for
all 10 entries across both JSON files, and added a regression test.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* test: improve assertion message per review feedback

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* test: add assertion for bare bedrock/moonshotai.kimi-k2.5 entry

Closes coverage gap identified by Greptile review - the bare
bedrock/moonshotai.kimi-k2.5 key (no region prefix) was fixed in
the JSON but not covered by any test assertion.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: RheagalFire <arishalam121@gmail.com>
This commit is contained in:
Ojasava Paras 2026-03-12 17:05:54 -04:00 • committed by GitHub
parent d34daeeda0
commit 9c01c08df2
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 76 additions and 43 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

40
litellm/model_prices_and_context_window_backup.json
View file

@ -6634,8 +6634,8 @@
"input_cost_per_token": 7.2e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3.6e-06,
"supports_function_calling": true,
@ -6672,8 +6672,8 @@
"input_cost_per_token": 6e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3.03e-06,
"source": "https://platform.moonshot.ai/docs/guide/kimi-k2-5-quickstart",
@ -6743,8 +6743,8 @@
"input_cost_per_token": 7.2e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3.6e-06,
"supports_function_calling": true,
@ -6796,8 +6796,8 @@
"input_cost_per_token": 7.2e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3.6e-06,
"supports_function_calling": true,
@ -6867,8 +6867,8 @@
"input_cost_per_token": 7.2e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3.6e-06,
"supports_function_calling": true,
@ -7211,8 +7211,8 @@
"input_cost_per_token": 7.2e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3.6e-06,
"supports_function_calling": true,
@ -7411,8 +7411,8 @@
"input_cost_per_token": 6e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3e-06,
"supports_function_calling": true,
@ -7475,8 +7475,8 @@
"input_cost_per_token": 6e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3e-06,
"supports_function_calling": true,
@ -7957,8 +7957,8 @@
"input_cost_per_token": 6e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3e-06,
"supports_function_calling": true,
@ -21900,8 +21900,8 @@
"input_cost_per_token": 6e-07,
"litellm_provider": "bedrock_converse",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3e-06,
"supports_function_calling": true,

40
model_prices_and_context_window.json
View file

@ -6634,8 +6634,8 @@
"input_cost_per_token": 7.2e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3.6e-06,
"supports_function_calling": true,
@ -6672,8 +6672,8 @@
"input_cost_per_token": 6e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3.03e-06,
"source": "https://platform.moonshot.ai/docs/guide/kimi-k2-5-quickstart",
@ -6743,8 +6743,8 @@
"input_cost_per_token": 7.2e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3.6e-06,
"supports_function_calling": true,
@ -6796,8 +6796,8 @@
"input_cost_per_token": 7.2e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3.6e-06,
"supports_function_calling": true,
@ -6867,8 +6867,8 @@
"input_cost_per_token": 7.2e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3.6e-06,
"supports_function_calling": true,
@ -7211,8 +7211,8 @@
"input_cost_per_token": 7.2e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3.6e-06,
"supports_function_calling": true,
@ -7411,8 +7411,8 @@
"input_cost_per_token": 6e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3e-06,
"supports_function_calling": true,
@ -7475,8 +7475,8 @@
"input_cost_per_token": 6e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3e-06,
"supports_function_calling": true,
@ -7957,8 +7957,8 @@
"input_cost_per_token": 6e-07,
"litellm_provider": "bedrock",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3e-06,
"supports_function_calling": true,
@ -21974,8 +21974,8 @@
"input_cost_per_token": 6e-07,
"litellm_provider": "bedrock_converse",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_token": 3e-06,
"supports_function_calling": true,

39
tests/litellm/test_bedrock_extended_beta_models.py
View file

@ -65,7 +65,7 @@ MODEL_CONFIGS = [
"us-west-2",
],
262144,
262144,
65536,
),
(
"qwen.qwen3-coder-next",
@ -158,10 +158,43 @@ class TestModelSpecificFeatures:
assert model_info["max_output_tokens"] == 8192
def test_moonshotai_kimi_k2_5_context_window(self):
"""Moonshot AI Kimi K2.5 has 256K context window"""
"""Moonshot AI Kimi K2.5 has 256K input, 64K output"""
model_info = get_model_info("bedrock/us-east-1/moonshotai.kimi-k2.5")
assert model_info["max_input_tokens"] == 262144
assert model_info["max_output_tokens"] == 262144
assert model_info["max_output_tokens"] == 65536
def test_moonshotai_kimi_k2_5_output_less_than_context_window(self):
"""Regression test for #22478: max_output_tokens must be < context window.
Bedrock enforces input_tokens + max_tokens <= context_window, so
max_output_tokens cannot equal the full 262144 context window.
"""
for region in [
"us-east-1",
"us-east-2",
"us-west-2",
"ap-northeast-1",
"ap-south-1",
]:
model_info = get_model_info(f"bedrock/{region}/moonshotai.kimi-k2.5")
assert model_info["max_output_tokens"] == 65536, (
f"bedrock/{region}/moonshotai.kimi-k2.5 max_output_tokens should be "
f"65536, got {model_info['max_output_tokens']}"
)
assert model_info["max_output_tokens"] < model_info["max_input_tokens"], (
f"max_output_tokens ({model_info['max_output_tokens']}) must be less "
f"than max_input_tokens ({model_info['max_input_tokens']}) to satisfy "
f"Bedrock constraint: input_tokens + max_tokens <= context_window"
)
# Also check the bare model entries
bare_bedrock_info = get_model_info("bedrock/moonshotai.kimi-k2.5")
assert bare_bedrock_info["max_output_tokens"] == 65536
assert bare_bedrock_info["max_input_tokens"] == 262144
bare_info = get_model_info("moonshotai.kimi-k2.5")
assert bare_info["max_output_tokens"] == 65536
assert bare_info["max_input_tokens"] == 262144
def test_qwen3_coder_next_context_window(self):
"""Qwen3 Coder Next has 256K input, 8K output"""