| # LiteLLM Proxy config for Replicate | |
| # Save as config.yaml and run: litellm --config ./config.yaml --debug | |
| model_list: | |
| # Meta Llama 3 Instruct (8B) via Replicate | |
| - model_name: openai/gpt-5-nano | |
| litellm_params: | |
| model: replicate/openai/gpt-5-nano | |
| api_key: os.environ/REPLICATE_API_KEY | |
| # optional common params | |
| temperature: 0.7 | |
| max_tokens: 1024 | |
| # Llama-2-70B chat with a pinned version ID (replace the example hash with the current one you use) | |
| - model_name: openai/gpt-5-mini | |
| litellm_params: | |
| model: replicate/openai/gpt-5-mini | |
| api_key: os.environ/REPLICATE_API_KEY | |
| # Call a Replicate *deployment* you own (replace owner/name) | |
| - model_name: openai/gpt-5 | |
| litellm_params: | |
| model: replicate/openai/gpt-5 | |
| api_key: os.environ/REPLICATE_API_KEY | |
| # Optional: proxy/router tuning | |
| router_settings: | |
| timeout: 600 # Replicate can cold-boot, so give it time | |
| num_retries: 1 | |
| # (Optional) enable simple logging callbacks, etc. | |
| # litellm_settings: | |
| # success_callback: ["generic"] | |