phoenix/litellm
1
0
Fork 0
forked from phoenix/litellm-mirror
Code Pull requests Activity

add cerebras cost tracking

Browse source
This commit is contained in:
Ishaan Jaff 2024-08-31 16:29:26 -07:00
parent fd4157cf71
commit 4bd85b9d83
2 changed files with 40 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

20
litellm/model_prices_and_context_window_backup.json
View file

@ -1273,6 +1273,26 @@
"mode": "chat", "mode": "chat",
"supports_function_calling": true "supports_function_calling": true
}, },
"cerebras/llama3.1-8b": {
"max_tokens": 128000,
"max_input_tokens": 128000,
"max_output_tokens": 128000,
"input_cost_per_token": 0.0000001,
"output_cost_per_token": 0.0000001,
"litellm_provider": "cerebras",
"mode": "chat",
"supports_function_calling": true
},
"cerebras/llama3.1-70b": {
"max_tokens": 128000,
"max_input_tokens": 128000,
"max_output_tokens": 128000,
"input_cost_per_token": 0.0000006,
"output_cost_per_token": 0.0000006,
"litellm_provider": "cerebras",
"mode": "chat",
"supports_function_calling": true
},
"friendliai/mixtral-8x7b-instruct-v0-1": { "friendliai/mixtral-8x7b-instruct-v0-1": {
"max_tokens": 32768, "max_tokens": 32768,
"max_input_tokens": 32768, "max_input_tokens": 32768,

20
model_prices_and_context_window.json
View file

@ -1273,6 +1273,26 @@
"mode": "chat", "mode": "chat",
"supports_function_calling": true "supports_function_calling": true
}, },
"cerebras/llama3.1-8b": {
"max_tokens": 128000,
"max_input_tokens": 128000,
"max_output_tokens": 128000,
"input_cost_per_token": 0.0000001,
"output_cost_per_token": 0.0000001,
"litellm_provider": "cerebras",
"mode": "chat",
"supports_function_calling": true
},
"cerebras/llama3.1-70b": {
"max_tokens": 128000,
"max_input_tokens": 128000,
"max_output_tokens": 128000,
"input_cost_per_token": 0.0000006,
"output_cost_per_token": 0.0000006,
"litellm_provider": "cerebras",
"mode": "chat",
"supports_function_calling": true
},
"friendliai/mixtral-8x7b-instruct-v0-1": { "friendliai/mixtral-8x7b-instruct-v0-1": {
"max_tokens": 32768, "max_tokens": 32768,
"max_input_tokens": 32768, "max_input_tokens": 32768,