scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-08 22:21:35 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 78

fix(model_prices): registry audit - add groq qwen3.8-27b, gemini live native audio (from #38573), anthropic sources

Browse source 
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
This commit is contained in:
Devin AI 2026-08-28 19:18:26 +00:00
parent 2ef77f30e3
commit 9df3a4fc1c
3 changed files with 143 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

67
litellm/model_prices_and_context_window_backup.json
View file

@ -12569,7 +12569,8 @@
"us": 1.1
},
"supports_output_config": true,
"prompt_cache_min_tokens": 1024
"prompt_cache_min_tokens": 1024,
"source": "https://docs.anthropic.com/en/docs/about-claude/models/overview"
},
"claude-sonnet-4-6": {
"deprecation_date": "2027-02-17",
@ -12971,7 +12972,8 @@
},
"supports_output_config": true,
"prompt_cache_min_tokens": 512,
"supports_native_structured_output": true
"supports_native_structured_output": true,
"source": "https://docs.anthropic.com/en/docs/about-claude/models/overview"
},
"claude-opus-5": {
"deprecation_date": "2027-07-24",
@ -13011,7 +13013,8 @@
},
"supports_output_config": true,
"supports_speed": true,
"prompt_cache_min_tokens": 512
"prompt_cache_min_tokens": 512,
"source": "https://docs.anthropic.com/en/docs/about-claude/models/overview"
},
"claude-opus-4-8": {
"deprecation_date": "2027-05-28",
@ -20536,7 +20539,51 @@
"output_cost_per_token": 2e-06,
"source": "https://cloud.google.com/vertex-ai/generative-ai/pricing",
"supported_endpoints": [
"/vertex_ai/live"
"/vertex_ai/live",
"/v1/realtime"
],
"supported_modalities": [
"text",
"image",
"audio",
"video"
],
"supported_output_modalities": [
"text",
"audio"
],
"supports_audio_input": true,
"supports_audio_output": true,
"supports_function_calling": true,
"supports_parallel_function_calling": true,
"supports_pdf_input": true,
"supports_prompt_caching": true,
"supports_response_schema": true,
"supports_system_messages": true,
"supports_tool_choice": true,
"supports_url_context": true,
"supports_vision": true,
"supports_web_search": true,
"search_context_cost_per_query": {
"search_context_size_low": 0.035,
"search_context_size_medium": 0.035,
"search_context_size_high": 0.035
},
"gemini_native_audio": true
},
"gemini/gemini-live-2.5-flash-native-audio": {
"input_cost_per_audio_token": 3e-06,
"input_cost_per_token": 5e-07,
"litellm_provider": "gemini",
"max_input_tokens": 1048576,
"max_output_tokens": 65535,
"max_tokens": 65535,
"mode": "realtime",
"output_cost_per_audio_token": 1.2e-05,
"output_cost_per_token": 2e-06,
"source": "https://ai.google.dev/gemini-api/docs/pricing",
"supported_endpoints": [
"/v1/realtime"
],
"supported_modalities": [
"text",
@ -54559,5 +54606,17 @@
"output_cost_per_token_above_200k_tokens": 5e-06,
"cache_read_input_token_cost_above_200k_tokens": 4e-07,
"supports_response_schema": true
},
"groq/qwen/qwen3.8-27b": {
"input_cost_per_token": 8e-07,
"litellm_provider": "groq",
"max_input_tokens": 131042,
"max_output_tokens": 16384,
"max_tokens": 16384,
"mode": "chat",
"output_cost_per_token": 4e-06,
"source": "https://console.groq.com/docs/model/qwen/qwen3.8-27b",
"supports_reasoning": true,
"supports_vision": true
}
}

67
model_prices_and_context_window.json
View file

@ -12569,7 +12569,8 @@
"us": 1.1
},
"supports_output_config": true,
"prompt_cache_min_tokens": 1024
"prompt_cache_min_tokens": 1024,
"source": "https://docs.anthropic.com/en/docs/about-claude/models/overview"
},
"claude-sonnet-4-6": {
"deprecation_date": "2027-02-17",
@ -12971,7 +12972,8 @@
},
"supports_output_config": true,
"prompt_cache_min_tokens": 512,
"supports_native_structured_output": true
"supports_native_structured_output": true,
"source": "https://docs.anthropic.com/en/docs/about-claude/models/overview"
},
"claude-opus-5": {
"deprecation_date": "2027-07-24",
@ -13011,7 +13013,8 @@
},
"supports_output_config": true,
"supports_speed": true,
"prompt_cache_min_tokens": 512
"prompt_cache_min_tokens": 512,
"source": "https://docs.anthropic.com/en/docs/about-claude/models/overview"
},
"claude-opus-4-8": {
"deprecation_date": "2027-05-28",
@ -20536,7 +20539,51 @@
"output_cost_per_token": 2e-06,
"source": "https://cloud.google.com/vertex-ai/generative-ai/pricing",
"supported_endpoints": [
"/vertex_ai/live"
"/vertex_ai/live",
"/v1/realtime"
],
"supported_modalities": [
"text",
"image",
"audio",
"video"
],
"supported_output_modalities": [
"text",
"audio"
],
"supports_audio_input": true,
"supports_audio_output": true,
"supports_function_calling": true,
"supports_parallel_function_calling": true,
"supports_pdf_input": true,
"supports_prompt_caching": true,
"supports_response_schema": true,
"supports_system_messages": true,
"supports_tool_choice": true,
"supports_url_context": true,
"supports_vision": true,
"supports_web_search": true,
"search_context_cost_per_query": {
"search_context_size_low": 0.035,
"search_context_size_medium": 0.035,
"search_context_size_high": 0.035
},
"gemini_native_audio": true
},
"gemini/gemini-live-2.5-flash-native-audio": {
"input_cost_per_audio_token": 3e-06,
"input_cost_per_token": 5e-07,
"litellm_provider": "gemini",
"max_input_tokens": 1048576,
"max_output_tokens": 65535,
"max_tokens": 65535,
"mode": "realtime",
"output_cost_per_audio_token": 1.2e-05,
"output_cost_per_token": 2e-06,
"source": "https://ai.google.dev/gemini-api/docs/pricing",
"supported_endpoints": [
"/v1/realtime"
],
"supported_modalities": [
"text",
@ -54559,5 +54606,17 @@
"output_cost_per_token_above_200k_tokens": 5e-06,
"cache_read_input_token_cost_above_200k_tokens": 4e-07,
"supports_response_schema": true
},
"groq/qwen/qwen3.8-27b": {
"input_cost_per_token": 8e-07,
"litellm_provider": "groq",
"max_input_tokens": 131042,
"max_output_tokens": 16384,
"max_tokens": 16384,
"mode": "chat",
"output_cost_per_token": 4e-06,
"source": "https://console.groq.com/docs/model/qwen/qwen3.8-27b",
"supports_reasoning": true,
"supports_vision": true
}
}

17
tests/test_litellm/llms/gemini/realtime/test_gemini_realtime_transformation.py
View file

@ -1812,6 +1812,8 @@ def patch_gemini_audio_cost_map_entries(monkeypatch):
"gemini-2.5-flash-native-audio",
"gemini-2.5-flash-native-audio-latest",
"gemini/gemini-2.5-flash-native-audio-latest",
"gemini-live-2.5-flash-native-audio",
"gemini/gemini-live-2.5-flash-native-audio",
]
flash_live_models = [
"gemini-3.1-flash-live-preview",
@ -1834,6 +1836,9 @@ def patch_gemini_audio_cost_map_entries(monkeypatch):
("gemini/gemini-3.1-flash-live-preview", True),
("gemini-2.5-flash-native-audio-latest", True),
("gemini/gemini-2.5-flash-native-audio-latest", True),
("gemini-live-2.5-flash-native-audio", True),
("gemini/gemini-live-2.5-flash-native-audio", True),
("vertex_ai/gemini-live-2.5-flash-native-audio", True),
("gemini-2.0-flash", False),
("gemini-2.5-flash", False),
],
@ -1842,6 +1847,18 @@ def test_is_audio_only_live_model_uses_cost_map(model, expected, patch_gemini_au
assert GeminiRealtimeConfig._is_audio_only_live_model(model) == expected
def test_gemini_live_native_audio_entries_in_catalog():
import json
from pathlib import Path
from typing import Final
catalog_path: Final = Path(__file__).parents[5] / "model_prices_and_context_window.json"
catalog: Final = json.loads(catalog_path.read_text())
for key in ("gemini-live-2.5-flash-native-audio", "gemini/gemini-live-2.5-flash-native-audio"):
assert key in catalog, f"{key} missing from model_prices_and_context_window.json"
assert catalog[key].get("gemini_native_audio") is True, f"{key}: gemini_native_audio must be true"
def test_is_setup_message_and_is_content_message():
config = GeminiRealtimeConfig()
assert config.is_setup_message({"setup": {}}) is True