scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-05 08:07:05 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 136

Merge pull request #38560 from BerriAI/devin/1787857843-registry-audit-rolling

Browse source 
fix(registry): add Gemini Omni 1.1 Flash, xAI grok-imagine image models, Mistral cache-read pricing, GLM 5.3 Flash + Kimi K2.7 Code entries
This commit is contained in:
Mateo Wang 2026-08-28 10:18:28 -07:00 • committed by GitHub
commit 5bcd494e88
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
5 changed files with 899 additions and 35 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

41
litellm/litellm_core_utils/llm_cost_calc/utils.py
View file

@ -1,6 +1,7 @@
# What is this?
## Helper utilities for cost_per_token()
import re
from collections.abc import Mapping
from dataclasses import dataclass
from types import MappingProxyType
@ -72,6 +73,19 @@ def _get_token_detail_value(details: object, key: str) -> int | None:
return value if isinstance(value, int) else None
_IMAGE_SIZE_PATTERN: Final = re.compile(r"\d+(?:x|-x-)\d+")
def _requested_image_param(optional_params: Mapping[str, object] | None, key: str) -> str | None:
value: Final = None if optional_params is None else optional_params.get(key)
return value if isinstance(value, str) else None
def _requested_image_size(optional_params: Mapping[str, object] | None) -> str | None:
value: Final = _requested_image_param(optional_params, "size")
return value if value is not None and _IMAGE_SIZE_PATTERN.fullmatch(value) else None
def get_web_search_requests(server_tool_use: Any) -> int | None:
"""
Tolerantly read ``web_search_requests`` from a ``server_tool_use`` value
@ -1311,12 +1325,13 @@ class CostCalculatorUtils:
cost_calculator as vertex_ai_image_cost_calculator,
)
if size is None:
size = completion_response.size or "1024-x-1024"
if quality is None:
quality = completion_response.quality or "standard"
if n is None:
n = len(completion_response.data) if completion_response.data else 0
resolved_size: Final = (
size or completion_response.size or _requested_image_size(optional_params) or "1024-x-1024"
)
resolved_quality: Final = (
quality or completion_response.quality or _requested_image_param(optional_params, "quality") or "standard"
)
resolved_n: Final = n if n is not None else (len(completion_response.data) if completion_response.data else 0)
if custom_llm_provider == litellm.LlmProviders.VERTEX_AI.value:
if isinstance(completion_response, ImageResponse):
@ -1328,7 +1343,7 @@ class CostCalculatorUtils:
if isinstance(completion_response, ImageResponse):
return bedrock_image_cost_calculator(
model=model,
size=size,
size=resolved_size,
image_response=completion_response,
optional_params=optional_params,
)
@ -1424,19 +1439,19 @@ class CostCalculatorUtils:
# Fall through to default for DALL-E models
return default_image_cost_calculator(
model=model,
quality=quality,
quality=resolved_quality,
custom_llm_provider=custom_llm_provider,
n=n,
size=size,
n=resolved_n,
size=resolved_size,
optional_params=optional_params,
)
else:
return default_image_cost_calculator(
model=model,
quality=quality,
quality=resolved_quality,
custom_llm_provider=custom_llm_provider,
n=n,
size=size,
n=resolved_n,
size=resolved_size,
optional_params=optional_params,
)
return 0.0

409
litellm/model_prices_and_context_window_backup.json
View file

@ -15216,6 +15216,29 @@
"supports_tool_choice": true,
"supports_vision": false
},
"databricks/databricks-glm-5-3-flash": {
"litellm_provider": "databricks",
"max_input_tokens": 1048576,
"max_output_tokens": 131072,
"max_tokens": 131072,
"metadata": {
"notes": "Databricks has not published pay-per-token DBU rates for this model yet (not on the foundation-model-serving pricing page as of 2026-08-27), so cost fields are omitted until rates are published."
},
"mode": "chat",
"source": "https://docs.databricks.com/aws/en/machine-learning/foundation-model-apis/supported-models",
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"text"
],
"supports_function_calling": true,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_tool_choice": true,
"supports_vision": true
},
"databricks/databricks-gpt-5": {
"cache_creation_input_token_cost": 1.24999e-06,
"cache_read_input_token_cost": 1.2502e-07,
@ -22971,9 +22994,9 @@
"input_cost_per_audio_token": 1.5e-06,
"input_cost_per_token": 1.5e-06,
"litellm_provider": "gemini",
"max_input_tokens": 1048576,
"max_output_tokens": 65535,
"max_tokens": 65535,
"max_input_tokens": 131072,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_reasoning_token": 9e-06,
"output_cost_per_token": 9e-06,
@ -22981,7 +23004,7 @@
"rpm": 2000,
"source": "https://ai.google.dev/gemini-api/docs/pricing",
"supported_endpoints": [
"/v1/chat/completions"
"/v1beta/interactions"
],
"supported_modalities": [
"text",
@ -30807,6 +30830,7 @@
"supports_tool_choice": true
},
"mistral/codestral-2508": {
"cache_read_input_token_cost": 3e-08,
"input_cost_per_token": 3e-07,
"litellm_provider": "mistral",
"max_input_tokens": 128000,
@ -30821,6 +30845,7 @@
"supports_tool_choice": true
},
"mistral/codestral-latest": {
"cache_read_input_token_cost": 3e-08,
"input_cost_per_token": 3e-07,
"litellm_provider": "mistral",
"max_input_tokens": 128000,
@ -30828,11 +30853,11 @@
"max_tokens": 128000,
"mode": "chat",
"output_cost_per_token": 9e-07,
"supports_assistant_prefill": true,
"supports_response_schema": true,
"supports_tool_choice": true,
"source": "https://docs.mistral.ai/models/model-cards/codestral-25-08",
"supports_function_calling": true
"supports_assistant_prefill": true,
"supports_function_calling": true,
"supports_response_schema": true,
"supports_tool_choice": true
},
"mistral/codestral-mamba-latest": {
"input_cost_per_token": 2.5e-07,
@ -31312,6 +31337,7 @@
"mode": "embedding"
},
"mistral/codestral-embed": {
"cache_read_input_token_cost": 1.5e-08,
"input_cost_per_token": 1.5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 8192,
@ -31319,6 +31345,7 @@
"mode": "embedding"
},
"mistral/codestral-embed-2505": {
"cache_read_input_token_cost": 1.5e-08,
"input_cost_per_token": 1.5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 8192,
@ -31368,6 +31395,7 @@
"supports_tool_choice": true
},
"mistral/mistral-large-latest": {
"cache_read_input_token_cost": 5e-08,
"input_cost_per_token": 5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31383,6 +31411,7 @@
"supports_vision": true
},
"mistral/mistral-large-3": {
"cache_read_input_token_cost": 5e-08,
"input_cost_per_token": 5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31398,6 +31427,7 @@
"supports_vision": true
},
"mistral/mistral-large-2512": {
"cache_read_input_token_cost": 5e-08,
"input_cost_per_token": 5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31468,6 +31498,7 @@
"supports_vision": true
},
"mistral/mistral-medium-2604": {
"cache_read_input_token_cost": 1.5e-07,
"input_cost_per_token": 1.5e-06,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31484,6 +31515,7 @@
"supports_vision": true
},
"mistral/mistral-medium-latest": {
"cache_read_input_token_cost": 1.5e-07,
"input_cost_per_token": 1.5e-06,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31516,6 +31548,7 @@
"supports_vision": true
},
"mistral/mistral-medium-3-5": {
"cache_read_input_token_cost": 1.5e-07,
"input_cost_per_token": 1.5e-06,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31545,6 +31578,7 @@
"supports_tool_choice": true
},
"mistral/mistral-small-latest": {
"cache_read_input_token_cost": 1.5e-08,
"input_cost_per_token": 1.5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31555,9 +31589,9 @@
"source": "https://docs.mistral.ai/models/model-cards/mistral-small-4-0-26-03",
"supports_assistant_prefill": true,
"supports_function_calling": true,
"supports_reasoning": true,
"supports_response_schema": true,
"supports_tool_choice": true,
"supports_reasoning": true,
"supports_vision": true
},
"mistral/mistral-small-3-2-2506": {
@ -31577,6 +31611,7 @@
"supports_vision": true
},
"mistral/ministral-3-3b-2512": {
"cache_read_input_token_cost": 1e-08,
"input_cost_per_token": 1e-07,
"litellm_provider": "mistral",
"max_input_tokens": 131072,
@ -31592,6 +31627,7 @@
"supports_vision": true
},
"mistral/ministral-3-8b-2512": {
"cache_read_input_token_cost": 1.5e-08,
"input_cost_per_token": 1.5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31607,6 +31643,7 @@
"supports_vision": true
},
"mistral/ministral-3-14b-2512": {
"cache_read_input_token_cost": 2e-08,
"input_cost_per_token": 2e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31622,6 +31659,7 @@
"supports_vision": true
},
"mistral/ministral-8b-2512": {
"cache_read_input_token_cost": 1.5e-08,
"input_cost_per_token": 1.5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31637,6 +31675,7 @@
"supports_vision": true
},
"mistral/ministral-8b-latest": {
"cache_read_input_token_cost": 1.5e-08,
"input_cost_per_token": 1.5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31844,6 +31883,24 @@
"supports_tool_choice": true,
"supports_web_search": true
},
"moonshot/kimi-k2.7-code": {
"cache_read_input_token_cost": 1.9e-07,
"input_cost_per_token": 9.5e-07,
"litellm_provider": "moonshot",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"mode": "chat",
"output_cost_per_token": 4e-06,
"source": "https://platform.kimi.ai/docs/pricing/chat-k27-code",
"supports_function_calling": true,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_response_schema": true,
"supports_tool_choice": true,
"supports_video_input": true,
"supports_vision": true
},
"moonshot/kimi-k2-turbo-preview": {
"cache_read_input_token_cost": 1.5e-07,
"deprecation_date": "2026-05-25",
@ -44611,6 +44668,22 @@
"supports_reasoning": true,
"supports_tool_choice": true
},
"zai/glm-5.3-flash": {
"cache_creation_input_token_cost": 0,
"cache_read_input_token_cost": 3e-08,
"input_cost_per_token": 1.5e-07,
"output_cost_per_token": 5e-07,
"litellm_provider": "zai",
"max_input_tokens": 1048576,
"max_output_tokens": 128000,
"mode": "chat",
"supports_function_calling": true,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_tool_choice": true,
"source": "https://docs.z.ai/guides/overview/pricing",
"supports_vision": true
},
"zai/glm-5.1": {
"cache_creation_input_token_cost": 0,
"cache_read_input_token_cost": 2.6e-07,
@ -51300,10 +51373,10 @@
"mode": "responses",
"output_cost_per_token": 2.5e-06,
"source": "https://docs.x.ai/docs/models",
"supports_function_calling": true,
"supports_function_calling": false,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_tool_choice": true,
"supports_tool_choice": false,
"supports_vision": true,
"supports_web_search": true,
"input_cost_per_token_above_200k_tokens": 2.5e-06,
@ -51497,6 +51570,7 @@
"web_search_billing_unit": "per_query"
},
"mistral/mistral-small-2603": {
"cache_read_input_token_cost": 1.5e-08,
"input_cost_per_token": 1.5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -54172,5 +54246,318 @@
"supports_reasoning": true,
"supports_vision": true,
"source": "https://deepinfra.com/pricing"
},
"gemini/gemini-omni-1.1-flash": {
"input_cost_per_audio_token": 1.5e-06,
"input_cost_per_token": 1.5e-06,
"litellm_provider": "gemini",
"max_input_tokens": 131072,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_reasoning_token": 9e-06,
"output_cost_per_token": 9e-06,
"output_cost_per_video_token": 1.75e-05,
"rpm": 2000,
"source": "https://ai.google.dev/gemini-api/docs/pricing",
"supported_endpoints": [
"/v1beta/interactions"
],
"supported_modalities": [
"text",
"image",
"audio",
"video"
],
"supported_output_modalities": [
"text",
"video"
],
"supports_audio_input": true,
"supports_reasoning": true,
"supports_system_messages": true,
"supports_video_input": true,
"supports_vision": true,
"tpm": 800000
},
"xai/grok-4.20": {
"cache_read_input_token_cost": 2e-07,
"input_cost_per_token": 1.25e-06,
"litellm_provider": "xai",
"max_input_tokens": 1000000,
"max_output_tokens": 1000000,
"max_tokens": 1000000,
"mode": "chat",
"output_cost_per_token": 2.5e-06,
"source": "https://docs.x.ai/docs/models",
"supports_function_calling": true,
"supports_reasoning": true,
"supports_tool_choice": true,
"supports_vision": true,
"supports_web_search": true,
"input_cost_per_token_above_200k_tokens": 2.5e-06,
"output_cost_per_token_above_200k_tokens": 5e-06,
"cache_read_input_token_cost_above_200k_tokens": 4e-07,
"supports_prompt_caching": true,
"supports_response_schema": true
},
"xai/grok-4.20-reasoning": {
"cache_read_input_token_cost": 2e-07,
"input_cost_per_token": 1.25e-06,
"litellm_provider": "xai",
"max_input_tokens": 1000000,
"max_output_tokens": 1000000,
"max_tokens": 1000000,
"mode": "chat",
"output_cost_per_token": 2.5e-06,
"source": "https://docs.x.ai/docs/models",
"supports_function_calling": true,
"supports_reasoning": true,
"supports_tool_choice": true,
"supports_vision": true,
"supports_web_search": true,
"input_cost_per_token_above_200k_tokens": 2.5e-06,
"output_cost_per_token_above_200k_tokens": 5e-06,
"cache_read_input_token_cost_above_200k_tokens": 4e-07,
"supports_prompt_caching": true,
"supports_response_schema": true
},
"xai/grok-4.20-reasoning-latest": {
"cache_read_input_token_cost": 2e-07,
"input_cost_per_token": 1.25e-06,
"litellm_provider": "xai",
"max_input_tokens": 1000000,
"max_output_tokens": 1000000,
"max_tokens": 1000000,
"mode": "chat",
"output_cost_per_token": 2.5e-06,
"source": "https://docs.x.ai/docs/models",
"supports_function_calling": true,
"supports_reasoning": true,
"supports_tool_choice": true,
"supports_vision": true,
"supports_web_search": true,
"input_cost_per_token_above_200k_tokens": 2.5e-06,
"output_cost_per_token_above_200k_tokens": 5e-06,
"cache_read_input_token_cost_above_200k_tokens": 4e-07,
"supports_prompt_caching": true,
"supports_response_schema": true
},
"xai/grok-imagine-image": {
"input_cost_per_image": 0.02,
"litellm_provider": "xai",
"mode": "image_generation",
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/images/generations"
],
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"image"
]
},
"xai/grok-imagine-image-2026-03-02": {
"input_cost_per_image": 0.02,
"litellm_provider": "xai",
"mode": "image_generation",
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/images/generations"
],
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"image"
]
},
"xai/grok-imagine-image-quality": {
"input_cost_per_image": 0.05,
"litellm_provider": "xai",
"mode": "image_generation",
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/images/generations"
],
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"image"
]
},
"xai/grok-imagine-image-quality-20260403": {
"input_cost_per_image": 0.05,
"litellm_provider": "xai",
"mode": "image_generation",
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/images/generations"
],
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"image"
]
},
"xai/grok-imagine-image-quality-latest": {
"input_cost_per_image": 0.05,
"litellm_provider": "xai",
"mode": "image_generation",
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/images/generations"
],
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"image"
]
},
"xai/grok-imagine-image-pro": {
"input_cost_per_image": 0.05,
"litellm_provider": "xai",
"mode": "image_generation",
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/images/generations"
],
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"image"
],
"deprecation_date": "2026-05-15"
},
"xai/grok-imagine-image-2.0": {
"input_cost_per_image": 0.06,
"litellm_provider": "xai",
"mode": "image_generation",
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/images/generations"
],
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"image"
]
},
"low/1024-x-1024/grok-imagine-image-2.0": {
"input_cost_per_image": 0.04,
"litellm_provider": "xai",
"mode": "image_generation",
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/images/generations"
],
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"image"
]
},
"xai/grok-4.20-non-reasoning": {
"cache_read_input_token_cost": 2e-07,
"input_cost_per_token": 1.25e-06,
"litellm_provider": "xai",
"max_input_tokens": 1000000,
"max_output_tokens": 1000000,
"max_tokens": 1000000,
"mode": "chat",
"output_cost_per_token": 2.5e-06,
"source": "https://docs.x.ai/docs/models",
"supports_function_calling": true,
"supports_prompt_caching": true,
"supports_tool_choice": true,
"supports_vision": true,
"supports_web_search": true,
"input_cost_per_token_above_200k_tokens": 2.5e-06,
"output_cost_per_token_above_200k_tokens": 5e-06,
"cache_read_input_token_cost_above_200k_tokens": 4e-07,
"supports_response_schema": true
},
"xai/grok-4.20-non-reasoning-latest": {
"cache_read_input_token_cost": 2e-07,
"input_cost_per_token": 1.25e-06,
"litellm_provider": "xai",
"max_input_tokens": 1000000,
"max_output_tokens": 1000000,
"max_tokens": 1000000,
"mode": "chat",
"output_cost_per_token": 2.5e-06,
"source": "https://docs.x.ai/docs/models",
"supports_function_calling": true,
"supports_prompt_caching": true,
"supports_tool_choice": true,
"supports_vision": true,
"supports_web_search": true,
"input_cost_per_token_above_200k_tokens": 2.5e-06,
"output_cost_per_token_above_200k_tokens": 5e-06,
"cache_read_input_token_cost_above_200k_tokens": 4e-07,
"supports_response_schema": true
},
"xai/grok-4.20-multi-agent": {
"cache_read_input_token_cost": 2e-07,
"input_cost_per_token": 1.25e-06,
"litellm_provider": "xai",
"max_input_tokens": 1000000,
"max_output_tokens": 1000000,
"max_tokens": 1000000,
"mode": "responses",
"output_cost_per_token": 2.5e-06,
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/responses"
],
"supports_function_calling": false,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_tool_choice": false,
"supports_vision": true,
"supports_web_search": true,
"input_cost_per_token_above_200k_tokens": 2.5e-06,
"output_cost_per_token_above_200k_tokens": 5e-06,
"cache_read_input_token_cost_above_200k_tokens": 4e-07,
"supports_response_schema": true
},
"xai/grok-4.20-multi-agent-latest": {
"cache_read_input_token_cost": 2e-07,
"input_cost_per_token": 1.25e-06,
"litellm_provider": "xai",
"max_input_tokens": 1000000,
"max_output_tokens": 1000000,
"max_tokens": 1000000,
"mode": "responses",
"output_cost_per_token": 2.5e-06,
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/responses"
],
"supports_function_calling": false,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_tool_choice": false,
"supports_vision": true,
"supports_web_search": true,
"input_cost_per_token_above_200k_tokens": 2.5e-06,
"output_cost_per_token_above_200k_tokens": 5e-06,
"cache_read_input_token_cost_above_200k_tokens": 4e-07,
"supports_response_schema": true
}
}

409
model_prices_and_context_window.json
View file

@ -15216,6 +15216,29 @@
"supports_tool_choice": true,
"supports_vision": false
},
"databricks/databricks-glm-5-3-flash": {
"litellm_provider": "databricks",
"max_input_tokens": 1048576,
"max_output_tokens": 131072,
"max_tokens": 131072,
"metadata": {
"notes": "Databricks has not published pay-per-token DBU rates for this model yet (not on the foundation-model-serving pricing page as of 2026-08-27), so cost fields are omitted until rates are published."
},
"mode": "chat",
"source": "https://docs.databricks.com/aws/en/machine-learning/foundation-model-apis/supported-models",
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"text"
],
"supports_function_calling": true,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_tool_choice": true,
"supports_vision": true
},
"databricks/databricks-gpt-5": {
"cache_creation_input_token_cost": 1.24999e-06,
"cache_read_input_token_cost": 1.2502e-07,
@ -22971,9 +22994,9 @@
"input_cost_per_audio_token": 1.5e-06,
"input_cost_per_token": 1.5e-06,
"litellm_provider": "gemini",
"max_input_tokens": 1048576,
"max_output_tokens": 65535,
"max_tokens": 65535,
"max_input_tokens": 131072,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_reasoning_token": 9e-06,
"output_cost_per_token": 9e-06,
@ -22981,7 +23004,7 @@
"rpm": 2000,
"source": "https://ai.google.dev/gemini-api/docs/pricing",
"supported_endpoints": [
"/v1/chat/completions"
"/v1beta/interactions"
],
"supported_modalities": [
"text",
@ -30807,6 +30830,7 @@
"supports_tool_choice": true
},
"mistral/codestral-2508": {
"cache_read_input_token_cost": 3e-08,
"input_cost_per_token": 3e-07,
"litellm_provider": "mistral",
"max_input_tokens": 128000,
@ -30821,6 +30845,7 @@
"supports_tool_choice": true
},
"mistral/codestral-latest": {
"cache_read_input_token_cost": 3e-08,
"input_cost_per_token": 3e-07,
"litellm_provider": "mistral",
"max_input_tokens": 128000,
@ -30828,11 +30853,11 @@
"max_tokens": 128000,
"mode": "chat",
"output_cost_per_token": 9e-07,
"supports_assistant_prefill": true,
"supports_response_schema": true,
"supports_tool_choice": true,
"source": "https://docs.mistral.ai/models/model-cards/codestral-25-08",
"supports_function_calling": true
"supports_assistant_prefill": true,
"supports_function_calling": true,
"supports_response_schema": true,
"supports_tool_choice": true
},
"mistral/codestral-mamba-latest": {
"input_cost_per_token": 2.5e-07,
@ -31312,6 +31337,7 @@
"mode": "embedding"
},
"mistral/codestral-embed": {
"cache_read_input_token_cost": 1.5e-08,
"input_cost_per_token": 1.5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 8192,
@ -31319,6 +31345,7 @@
"mode": "embedding"
},
"mistral/codestral-embed-2505": {
"cache_read_input_token_cost": 1.5e-08,
"input_cost_per_token": 1.5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 8192,
@ -31368,6 +31395,7 @@
"supports_tool_choice": true
},
"mistral/mistral-large-latest": {
"cache_read_input_token_cost": 5e-08,
"input_cost_per_token": 5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31383,6 +31411,7 @@
"supports_vision": true
},
"mistral/mistral-large-3": {
"cache_read_input_token_cost": 5e-08,
"input_cost_per_token": 5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31398,6 +31427,7 @@
"supports_vision": true
},
"mistral/mistral-large-2512": {
"cache_read_input_token_cost": 5e-08,
"input_cost_per_token": 5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31468,6 +31498,7 @@
"supports_vision": true
},
"mistral/mistral-medium-2604": {
"cache_read_input_token_cost": 1.5e-07,
"input_cost_per_token": 1.5e-06,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31484,6 +31515,7 @@
"supports_vision": true
},
"mistral/mistral-medium-latest": {
"cache_read_input_token_cost": 1.5e-07,
"input_cost_per_token": 1.5e-06,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31516,6 +31548,7 @@
"supports_vision": true
},
"mistral/mistral-medium-3-5": {
"cache_read_input_token_cost": 1.5e-07,
"input_cost_per_token": 1.5e-06,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31545,6 +31578,7 @@
"supports_tool_choice": true
},
"mistral/mistral-small-latest": {
"cache_read_input_token_cost": 1.5e-08,
"input_cost_per_token": 1.5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31555,9 +31589,9 @@
"source": "https://docs.mistral.ai/models/model-cards/mistral-small-4-0-26-03",
"supports_assistant_prefill": true,
"supports_function_calling": true,
"supports_reasoning": true,
"supports_response_schema": true,
"supports_tool_choice": true,
"supports_reasoning": true,
"supports_vision": true
},
"mistral/mistral-small-3-2-2506": {
@ -31577,6 +31611,7 @@
"supports_vision": true
},
"mistral/ministral-3-3b-2512": {
"cache_read_input_token_cost": 1e-08,
"input_cost_per_token": 1e-07,
"litellm_provider": "mistral",
"max_input_tokens": 131072,
@ -31592,6 +31627,7 @@
"supports_vision": true
},
"mistral/ministral-3-8b-2512": {
"cache_read_input_token_cost": 1.5e-08,
"input_cost_per_token": 1.5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31607,6 +31643,7 @@
"supports_vision": true
},
"mistral/ministral-3-14b-2512": {
"cache_read_input_token_cost": 2e-08,
"input_cost_per_token": 2e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31622,6 +31659,7 @@
"supports_vision": true
},
"mistral/ministral-8b-2512": {
"cache_read_input_token_cost": 1.5e-08,
"input_cost_per_token": 1.5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31637,6 +31675,7 @@
"supports_vision": true
},
"mistral/ministral-8b-latest": {
"cache_read_input_token_cost": 1.5e-08,
"input_cost_per_token": 1.5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -31844,6 +31883,24 @@
"supports_tool_choice": true,
"supports_web_search": true
},
"moonshot/kimi-k2.7-code": {
"cache_read_input_token_cost": 1.9e-07,
"input_cost_per_token": 9.5e-07,
"litellm_provider": "moonshot",
"max_input_tokens": 262144,
"max_output_tokens": 262144,
"max_tokens": 262144,
"mode": "chat",
"output_cost_per_token": 4e-06,
"source": "https://platform.kimi.ai/docs/pricing/chat-k27-code",
"supports_function_calling": true,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_response_schema": true,
"supports_tool_choice": true,
"supports_video_input": true,
"supports_vision": true
},
"moonshot/kimi-k2-turbo-preview": {
"cache_read_input_token_cost": 1.5e-07,
"deprecation_date": "2026-05-25",
@ -44611,6 +44668,22 @@
"supports_reasoning": true,
"supports_tool_choice": true
},
"zai/glm-5.3-flash": {
"cache_creation_input_token_cost": 0,
"cache_read_input_token_cost": 3e-08,
"input_cost_per_token": 1.5e-07,
"output_cost_per_token": 5e-07,
"litellm_provider": "zai",
"max_input_tokens": 1048576,
"max_output_tokens": 128000,
"mode": "chat",
"supports_function_calling": true,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_tool_choice": true,
"source": "https://docs.z.ai/guides/overview/pricing",
"supports_vision": true
},
"zai/glm-5.1": {
"cache_creation_input_token_cost": 0,
"cache_read_input_token_cost": 2.6e-07,
@ -51300,10 +51373,10 @@
"mode": "responses",
"output_cost_per_token": 2.5e-06,
"source": "https://docs.x.ai/docs/models",
"supports_function_calling": true,
"supports_function_calling": false,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_tool_choice": true,
"supports_tool_choice": false,
"supports_vision": true,
"supports_web_search": true,
"input_cost_per_token_above_200k_tokens": 2.5e-06,
@ -51497,6 +51570,7 @@
"web_search_billing_unit": "per_query"
},
"mistral/mistral-small-2603": {
"cache_read_input_token_cost": 1.5e-08,
"input_cost_per_token": 1.5e-07,
"litellm_provider": "mistral",
"max_input_tokens": 262144,
@ -54172,5 +54246,318 @@
"supports_reasoning": true,
"supports_vision": true,
"source": "https://deepinfra.com/pricing"
},
"gemini/gemini-omni-1.1-flash": {
"input_cost_per_audio_token": 1.5e-06,
"input_cost_per_token": 1.5e-06,
"litellm_provider": "gemini",
"max_input_tokens": 131072,
"max_output_tokens": 65536,
"max_tokens": 65536,
"mode": "chat",
"output_cost_per_reasoning_token": 9e-06,
"output_cost_per_token": 9e-06,
"output_cost_per_video_token": 1.75e-05,
"rpm": 2000,
"source": "https://ai.google.dev/gemini-api/docs/pricing",
"supported_endpoints": [
"/v1beta/interactions"
],
"supported_modalities": [
"text",
"image",
"audio",
"video"
],
"supported_output_modalities": [
"text",
"video"
],
"supports_audio_input": true,
"supports_reasoning": true,
"supports_system_messages": true,
"supports_video_input": true,
"supports_vision": true,
"tpm": 800000
},
"xai/grok-4.20": {
"cache_read_input_token_cost": 2e-07,
"input_cost_per_token": 1.25e-06,
"litellm_provider": "xai",
"max_input_tokens": 1000000,
"max_output_tokens": 1000000,
"max_tokens": 1000000,
"mode": "chat",
"output_cost_per_token": 2.5e-06,
"source": "https://docs.x.ai/docs/models",
"supports_function_calling": true,
"supports_reasoning": true,
"supports_tool_choice": true,
"supports_vision": true,
"supports_web_search": true,
"input_cost_per_token_above_200k_tokens": 2.5e-06,
"output_cost_per_token_above_200k_tokens": 5e-06,
"cache_read_input_token_cost_above_200k_tokens": 4e-07,
"supports_prompt_caching": true,
"supports_response_schema": true
},
"xai/grok-4.20-reasoning": {
"cache_read_input_token_cost": 2e-07,
"input_cost_per_token": 1.25e-06,
"litellm_provider": "xai",
"max_input_tokens": 1000000,
"max_output_tokens": 1000000,
"max_tokens": 1000000,
"mode": "chat",
"output_cost_per_token": 2.5e-06,
"source": "https://docs.x.ai/docs/models",
"supports_function_calling": true,
"supports_reasoning": true,
"supports_tool_choice": true,
"supports_vision": true,
"supports_web_search": true,
"input_cost_per_token_above_200k_tokens": 2.5e-06,
"output_cost_per_token_above_200k_tokens": 5e-06,
"cache_read_input_token_cost_above_200k_tokens": 4e-07,
"supports_prompt_caching": true,
"supports_response_schema": true
},
"xai/grok-4.20-reasoning-latest": {
"cache_read_input_token_cost": 2e-07,
"input_cost_per_token": 1.25e-06,
"litellm_provider": "xai",
"max_input_tokens": 1000000,
"max_output_tokens": 1000000,
"max_tokens": 1000000,
"mode": "chat",
"output_cost_per_token": 2.5e-06,
"source": "https://docs.x.ai/docs/models",
"supports_function_calling": true,
"supports_reasoning": true,
"supports_tool_choice": true,
"supports_vision": true,
"supports_web_search": true,
"input_cost_per_token_above_200k_tokens": 2.5e-06,
"output_cost_per_token_above_200k_tokens": 5e-06,
"cache_read_input_token_cost_above_200k_tokens": 4e-07,
"supports_prompt_caching": true,
"supports_response_schema": true
},
"xai/grok-imagine-image": {
"input_cost_per_image": 0.02,
"litellm_provider": "xai",
"mode": "image_generation",
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/images/generations"
],
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"image"
]
},
"xai/grok-imagine-image-2026-03-02": {
"input_cost_per_image": 0.02,
"litellm_provider": "xai",
"mode": "image_generation",
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/images/generations"
],
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"image"
]
},
"xai/grok-imagine-image-quality": {
"input_cost_per_image": 0.05,
"litellm_provider": "xai",
"mode": "image_generation",
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/images/generations"
],
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"image"
]
},
"xai/grok-imagine-image-quality-20260403": {
"input_cost_per_image": 0.05,
"litellm_provider": "xai",
"mode": "image_generation",
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/images/generations"
],
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"image"
]
},
"xai/grok-imagine-image-quality-latest": {
"input_cost_per_image": 0.05,
"litellm_provider": "xai",
"mode": "image_generation",
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/images/generations"
],
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"image"
]
},
"xai/grok-imagine-image-pro": {
"input_cost_per_image": 0.05,
"litellm_provider": "xai",
"mode": "image_generation",
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/images/generations"
],
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"image"
],
"deprecation_date": "2026-05-15"
},
"xai/grok-imagine-image-2.0": {
"input_cost_per_image": 0.06,
"litellm_provider": "xai",
"mode": "image_generation",
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/images/generations"
],
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"image"
]
},
"low/1024-x-1024/grok-imagine-image-2.0": {
"input_cost_per_image": 0.04,
"litellm_provider": "xai",
"mode": "image_generation",
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/images/generations"
],
"supported_modalities": [
"text",
"image"
],
"supported_output_modalities": [
"image"
]
},
"xai/grok-4.20-non-reasoning": {
"cache_read_input_token_cost": 2e-07,
"input_cost_per_token": 1.25e-06,
"litellm_provider": "xai",
"max_input_tokens": 1000000,
"max_output_tokens": 1000000,
"max_tokens": 1000000,
"mode": "chat",
"output_cost_per_token": 2.5e-06,
"source": "https://docs.x.ai/docs/models",
"supports_function_calling": true,
"supports_prompt_caching": true,
"supports_tool_choice": true,
"supports_vision": true,
"supports_web_search": true,
"input_cost_per_token_above_200k_tokens": 2.5e-06,
"output_cost_per_token_above_200k_tokens": 5e-06,
"cache_read_input_token_cost_above_200k_tokens": 4e-07,
"supports_response_schema": true
},
"xai/grok-4.20-non-reasoning-latest": {
"cache_read_input_token_cost": 2e-07,
"input_cost_per_token": 1.25e-06,
"litellm_provider": "xai",
"max_input_tokens": 1000000,
"max_output_tokens": 1000000,
"max_tokens": 1000000,
"mode": "chat",
"output_cost_per_token": 2.5e-06,
"source": "https://docs.x.ai/docs/models",
"supports_function_calling": true,
"supports_prompt_caching": true,
"supports_tool_choice": true,
"supports_vision": true,
"supports_web_search": true,
"input_cost_per_token_above_200k_tokens": 2.5e-06,
"output_cost_per_token_above_200k_tokens": 5e-06,
"cache_read_input_token_cost_above_200k_tokens": 4e-07,
"supports_response_schema": true
},
"xai/grok-4.20-multi-agent": {
"cache_read_input_token_cost": 2e-07,
"input_cost_per_token": 1.25e-06,
"litellm_provider": "xai",
"max_input_tokens": 1000000,
"max_output_tokens": 1000000,
"max_tokens": 1000000,
"mode": "responses",
"output_cost_per_token": 2.5e-06,
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/responses"
],
"supports_function_calling": false,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_tool_choice": false,
"supports_vision": true,
"supports_web_search": true,
"input_cost_per_token_above_200k_tokens": 2.5e-06,
"output_cost_per_token_above_200k_tokens": 5e-06,
"cache_read_input_token_cost_above_200k_tokens": 4e-07,
"supports_response_schema": true
},
"xai/grok-4.20-multi-agent-latest": {
"cache_read_input_token_cost": 2e-07,
"input_cost_per_token": 1.25e-06,
"litellm_provider": "xai",
"max_input_tokens": 1000000,
"max_output_tokens": 1000000,
"max_tokens": 1000000,
"mode": "responses",
"output_cost_per_token": 2.5e-06,
"source": "https://docs.x.ai/docs/models",
"supported_endpoints": [
"/v1/responses"
],
"supports_function_calling": false,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_tool_choice": false,
"supports_vision": true,
"supports_web_search": true,
"input_cost_per_token_above_200k_tokens": 2.5e-06,
"output_cost_per_token_above_200k_tokens": 5e-06,
"cache_read_input_token_cost_above_200k_tokens": 4e-07,
"supports_response_schema": true
}
}

74
tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py
View file

@ -27,6 +27,7 @@ from litellm.types.utils import (
)
from litellm.litellm_core_utils.llm_cost_calc.utils import (
CostCalculatorUtils,
PromptTokensDetailsResult,
TokenTypeCostBreakdown,
_calculate_input_cost,
@ -3872,3 +3873,76 @@ def test_generic_cost_per_token_grok_46_long_context(_local_model_cost_map):
)
assert prompt_cost == pytest.approx(200_000 * 4e-06 + 50_000 * 1e-06)
assert completion_cost == pytest.approx(1_000 * 1.2e-05)
@pytest.mark.parametrize(
("response_quality", "requested_quality", "expected_cost"),
[
(None, "low", 0.04),
(None, None, 0.06),
("high", "low", 0.08),
],
)
def test_route_image_generation_cost_falls_back_to_requested_quality(
monkeypatch, response_quality, requested_quality, expected_cost
):
def tier(cost):
return {"litellm_provider": "xai", "mode": "image_generation", "input_cost_per_image": cost}
monkeypatch.setattr(
litellm,
"model_cost",
{
"xai/grok-imagine-image-2.0": tier(0.06),
"low/1024-x-1024/grok-imagine-image-2.0": tier(0.04),
"high/1024-x-1024/grok-imagine-image-2.0": tier(0.08),
},
)
response = ImageResponse(data=[ImageObject(url="https://example.com/image.png")], quality=response_quality)
optional_params = {} if requested_quality is None else {"quality": requested_quality}
cost = CostCalculatorUtils.route_image_generation_cost_calculator(
model="xai/grok-imagine-image-2.0",
completion_response=response,
custom_llm_provider="xai",
optional_params=optional_params,
call_type="image_generation",
)
assert cost == expected_cost
@pytest.mark.parametrize(
("requested_size", "expected_cost"),
[
("1536x1024", 0.05),
("1536-x-1024", 0.05),
("auto", 0.04),
(None, 0.04),
],
)
def test_route_image_generation_cost_falls_back_to_requested_size(monkeypatch, requested_size, expected_cost):
def tier(cost):
return {"litellm_provider": "xai", "mode": "image_generation", "input_cost_per_image": cost}
monkeypatch.setattr(
litellm,
"model_cost",
{
"xai/grok-imagine-image-2.0": tier(0.06),
"low/1024-x-1024/grok-imagine-image-2.0": tier(0.04),
"low/1536-x-1024/grok-imagine-image-2.0": tier(0.05),
},
)
response = ImageResponse(data=[ImageObject(url="https://example.com/image.png")])
optional_params = {"quality": "low", **({} if requested_size is None else {"size": requested_size})}
cost = CostCalculatorUtils.route_image_generation_cost_calculator(
model="xai/grok-imagine-image-2.0",
completion_response=response,
custom_llm_provider="xai",
optional_params=optional_params,
call_type="image_generation",
)
assert cost == expected_cost

1
tests/test_litellm/test_utils.py
View file

@ -1051,6 +1051,7 @@ def test_aaamodel_prices_and_context_window_json_is_valid():
"/v1/audio/speech",
"/v1/ocr",
"/vertex_ai/live",
"/v1beta/interactions",
],
},
},