diff options
| author | sigoden <sigoden@gmail.com> | 2024-07-25 15:56:56 -0700 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-07-26 06:56:56 +0800 |
| commit | 96ad64352dccfb9328374ef754c2a90a31c0a339 (patch) | |
| tree | 16132de380950739e22dd126e34d5495826a9192 /models.yaml | |
| parent | 54e78583991918c05e724ed70c2c4be4a15bc849 (diff) | |
| download | aichat-96ad64352dccfb9328374ef754c2a90a31c0a339.tar.gz | |
feat: merge vertexai-cluade with vertexai (#745)
Diffstat (limited to 'models.yaml')
| -rw-r--r-- | models.yaml | 36 |
1 files changed, 15 insertions, 21 deletions
diff --git a/models.yaml b/models.yaml index 0df9944..cdced59 100644 --- a/models.yaml +++ b/models.yaml @@ -277,6 +277,7 @@ - platform: vertexai # docs: # - https://cloud.google.com/vertex-ai/generative-ai/docs/learn/models + # - https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models/use-claude # - https://cloud.google.com/vertex-ai/generative-ai/pricing # - https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/gemini # notes: @@ -302,27 +303,6 @@ input_price: 0.125 output_price: 0.375 supports_function_calling: true - - name: text-embedding-004 - type: embedding - max_input_tokens: 3072 - input_price: 0.025 - output_vector_size: 768 - default_chunk_size: 1500 - max_batch_size: 5 - - name: text-multilingual-embedding-002 - type: embedding - max_input_tokens: 3072 - input_price: 0.2 - output_vector_size: 768 - default_chunk_size: 1500 - max_batch_size: 5 - -- platform: vertexai-claude - # docs: - # - https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models/use-claude - # notes: - # - get max_output_tokens info from models doc - models: - name: claude-3-5-sonnet@20240620 max_input_tokens: 200000 max_output_tokens: 4096 @@ -355,6 +335,20 @@ output_price: 1.25 supports_vision: true supports_function_calling: true + - name: text-embedding-004 + type: embedding + max_input_tokens: 3072 + input_price: 0.025 + output_vector_size: 768 + default_chunk_size: 1500 + max_batch_size: 5 + - name: text-multilingual-embedding-002 + type: embedding + max_input_tokens: 3072 + input_price: 0.2 + output_vector_size: 768 + default_chunk_size: 1500 + max_batch_size: 5 - platform: bedrock # docs: |