diff options
Diffstat (limited to 'models.yaml')
| -rw-r--r-- | models.yaml | 228 |
1 files changed, 186 insertions, 42 deletions
diff --git a/models.yaml b/models.yaml index 1a426ed..7e2a03e 100644 --- a/models.yaml +++ b/models.yaml @@ -3,6 +3,34 @@ # - https://platform.openai.com/docs/api-reference/chat - provider: openai models: + - name: gpt-5 + max_input_tokens: 400000 + max_output_tokens: 128000 + input_price: 1.25 + output_price: 10 + supports_vision: true + supports_function_calling: true + - name: gpt-5-chat-latest + max_input_tokens: 400000 + max_output_tokens: 128000 + input_price: 1.25 + output_price: 10 + supports_vision: true + supports_function_calling: true + - name: gpt-5-mini + max_input_tokens: 400000 + max_output_tokens: 128000 + input_price: 0.25 + output_price: 2 + supports_vision: true + supports_function_calling: true + - name: gpt-5-nano + max_input_tokens: 400000 + max_output_tokens: 128000 + input_price: 0.05 + output_price: 0.4 + supports_vision: true + supports_function_calling: true - name: gpt-4.1 max_input_tokens: 1047576 max_output_tokens: 32768 @@ -200,6 +228,30 @@ # - https://docs.anthropic.com/en/api/messages - provider: claude models: + - name: claude-opus-4-1-20250805 + max_input_tokens: 200000 + max_output_tokens: 8192 + require_max_tokens: true + input_price: 15 + output_price: 75 + supports_vision: true + supports_function_calling: true + - name: claude-opus-4-1-20250805:thinking + real_name: claude-opus-4-1-20250805 + max_input_tokens: 200000 + max_output_tokens: 8192 + require_max_tokens: true + input_price: 15 + output_price: 75 + supports_vision: true + supports_function_calling: true + patch: + body: + temperature: null + top_p: null + thinking: + type: enabled + budget_tokens: 16000 - name: claude-opus-4-20250514 max_input_tokens: 200000 max_output_tokens: 8192 @@ -271,14 +323,6 @@ thinking: type: enabled budget_tokens: 16000 - - name: claude-3-5-sonnet-20241022 - max_input_tokens: 200000 - max_output_tokens: 8192 - require_max_tokens: true - input_price: 3 - output_price: 15 - supports_vision: true - supports_function_calling: true - name: claude-3-5-haiku-20241022 max_input_tokens: 200000 max_output_tokens: 8192 @@ -466,6 +510,16 @@ # - https://console.groq.com/docs/api-reference#chat - provider: groq models: + - name: openai/gpt-oss-120b + max_input_tokens: 131072 + input_price: 0 + output_price: 0 + supports_function_calling: true + - name: openai/gpt-oss-20b + max_input_tokens: 131072 + input_price: 0 + output_price: 0 + supports_function_calling: true - name: meta-llama/llama-4-maverick-17b-128e-instruct max_input_tokens: 131072 input_price: 0 @@ -542,6 +596,29 @@ output_price: 0.3 supports_vision: true supports_function_calling: true + - name: claude-opus-4-1@20250805 + max_input_tokens: 200000 + max_output_tokens: 8192 + require_max_tokens: true + input_price: 15 + output_price: 75 + supports_vision: true + supports_function_calling: true + - name: claude-opus-4-1@20250805:thinking + real_name: claude-opus-4-1@20250805 + max_input_tokens: 200000 + max_output_tokens: 24000 + require_max_tokens: true + input_price: 15 + output_price: 75 + supports_vision: true + patch: + body: + temperature: null + top_p: null + thinking: + type: enabled + budget_tokens: 16000 - name: claude-opus-4@20250514 max_input_tokens: 200000 max_output_tokens: 8192 @@ -611,14 +688,6 @@ thinking: type: enabled budget_tokens: 16000 - - name: claude-3-5-sonnet-v2@20241022 - max_input_tokens: 200000 - max_output_tokens: 8192 - require_max_tokens: true - input_price: 3 - output_price: 15 - supports_vision: true - supports_function_calling: true - name: claude-3-5-haiku@20241022 max_input_tokens: 200000 max_output_tokens: 8192 @@ -659,6 +728,31 @@ # - https://docs.aws.amazon.com/bedrock/latest/userguide/conversation-inference-call.html - provider: bedrock models: + - name: us.anthropic.claude-opus-4-1-20250805-v1:0 + max_input_tokens: 200000 + max_output_tokens: 8192 + require_max_tokens: true + input_price: 15 + output_price: 75 + supports_vision: true + supports_function_calling: true + - name: us.anthropic.claude-opus-4-1-20250805-v1:0:thinking + real_name: us.anthropic.claude-opus-4-1-20250805-v1:0 + max_input_tokens: 200000 + max_output_tokens: 24000 + require_max_tokens: true + input_price: 15 + output_price: 75 + supports_vision: true + patch: + body: + inferenceConfig: + temperature: null + topP: null + additionalModelRequestFields: + thinking: + type: enabled + budget_tokens: 16000 - name: us.anthropic.claude-opus-4-20250514-v1:0 max_input_tokens: 200000 max_output_tokens: 8192 @@ -734,14 +828,6 @@ thinking: type: enabled budget_tokens: 16000 - - name: anthropic.claude-3-5-sonnet-20241022-v2:0 - max_input_tokens: 200000 - max_output_tokens: 8192 - require_max_tokens: true - input_price: 3 - output_price: 15 - supports_vision: true - supports_function_calling: true - name: anthropic.claude-3-5-haiku-20241022-v1:0 max_input_tokens: 200000 max_output_tokens: 8192 @@ -1185,6 +1271,34 @@ # - https://openrouter.ai/docs/api-reference/chat-completion - provider: openrouter models: + - name: openai/gpt-5 + max_input_tokens: 400000 + max_output_tokens: 128000 + input_price: 1.25 + output_price: 10 + supports_vision: true + supports_function_calling: true + - name: openai/gpt-5-chat + max_input_tokens: 400000 + max_output_tokens: 128000 + input_price: 1.25 + output_price: 10 + supports_vision: true + supports_function_calling: true + - name: openai/gpt-5-mini + max_input_tokens: 400000 + max_output_tokens: 128000 + input_price: 0.25 + output_price: 2 + supports_vision: true + supports_function_calling: true + - name: openai/gpt-5-nano + max_input_tokens: 400000 + max_output_tokens: 128000 + input_price: 0.05 + output_price: 0.4 + supports_vision: true + supports_function_calling: true - name: openai/gpt-4.1 max_input_tokens: 1047576 max_output_tokens: 32768 @@ -1290,6 +1404,16 @@ body: temperature: null top_p: null + - name: openai/gpt-oss-120b + max_input_tokens: 131072 + input_price: 0.09 + output_price: 0.45 + supports_function_calling: true + - name: openai/gpt-oss-20b + max_input_tokens: 131072 + input_price: 0.04 + output_price: 0.16 + supports_function_calling: true - name: google/gemini-2.5-flash max_input_tokens: 1048576 input_price: 0.3 @@ -1323,7 +1447,7 @@ max_input_tokens: 131072 input_price: 0.1 output_price: 0.2 - - name: anthropic/claude-opus-4 + - name: anthropic/claude-opus-4.1 max_input_tokens: 200000 max_output_tokens: 8192 require_max_tokens: true @@ -1331,15 +1455,15 @@ output_price: 75 supports_vision: true supports_function_calling: true - - name: anthropic/claude-sonnet-4 + - name: anthropic/claude-opus-4 max_input_tokens: 200000 max_output_tokens: 8192 require_max_tokens: true - input_price: 3 - output_price: 15 + input_price: 15 + output_price: 75 supports_vision: true supports_function_calling: true - - name: anthropic/claude-3.7-sonnet + - name: anthropic/claude-sonnet-4 max_input_tokens: 200000 max_output_tokens: 8192 require_max_tokens: true @@ -1347,17 +1471,7 @@ output_price: 15 supports_vision: true supports_function_calling: true - - name: anthropic/claude-3.7-sonnet:thinking - max_input_tokens: 200000 - max_output_tokens: 24000 - require_max_tokens: true - input_price: 3 - output_price: 15 - supports_vision: true - patch: - body: - include_reasoning: true - - name: anthropic/claude-3.5-sonnet + - name: anthropic/claude-3.7-sonnet max_input_tokens: 200000 max_output_tokens: 8192 require_max_tokens: true @@ -1427,12 +1541,12 @@ input_price: 0.3 output_price: 0.9 supports_function_calling: true - - name: ai21/jamba-1.6-large + - name: ai21/jamba-large-1.7 max_input_tokens: 256000 input_price: 2 output_price: 8 supports_function_calling: true - - name: ai21/jamba-1.6-mini + - name: ai21/jamba-mini-1.7 max_input_tokens: 256000 input_price: 0.2 output_price: 0.4 @@ -1609,6 +1723,26 @@ # - https://github.com/marketplace?type=models - provider: github models: + - name: gpt-5 + max_input_tokens: 400000 + max_output_tokens: 128000 + supports_vision: true + supports_function_calling: true + - name: gpt-5-chat + max_input_tokens: 400000 + max_output_tokens: 128000 + supports_vision: true + supports_function_calling: true + - name: gpt-5-mini + max_input_tokens: 400000 + max_output_tokens: 128000 + supports_vision: true + supports_function_calling: true + - name: gpt-5-nano + max_input_tokens: 400000 + max_output_tokens: 128000 + supports_vision: true + supports_function_calling: true - name: gpt-4.1 max_input_tokens: 1047576 max_output_tokens: 32768 @@ -1759,6 +1893,16 @@ # - https://deepinfra.com/docs/openai_api - provider: deepinfra models: + - name: openai/gpt-oss-120b + max_input_tokens: 131072 + input_price: 0.09 + output_price: 0.45 + supports_function_calling: true + - name: openai/gpt-oss-20b + max_input_tokens: 131072 + input_price: 0.04 + output_price: 0.16 + supports_function_calling: true - name: meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8 max_input_tokens: 1048576 input_price: 0.18 |