Here is the full list of the supported models:

DeepSeek R1

• 8-bit: DeepSeek-R1-Turbo

LlaMA-3.3 70B Instruct

• 8-bit: Meta-Llama-3.3-70B-Instruct-Turbo

LLaMA-3.1 8B Instruct

• 16-bit: Meta-Llama-3.1-8B-Instruct-Reference

• 8-bit: Meta-Llama-3.1-8B-Instruct-Turbo

• 4-bit: Meta-Llama-3.1-8B-Instruct-Lite

LLaMA-3.1 70B Instruct

• 16-bit: Meta-Llama-3.1-70B-Instruct-Reference

• 8-bit: Meta-Llama-3.1-70B-Instruct-Turbo

• 4-bit: Meta-Llama-3.1-70B-Instruct-Lite

LLaMA-3.1 405B Instruct

• 8-bit: Meta-Llama-3.1-405B-Instruct-Turbo

• 4-bit: Meta-Llama-3.1-405B-Instruct-Lite

An updated list of supported models can be found by calling the /v1/models endpoint.

We are continuously working to expand our model offerings and will update this list as new models become available. Stay tuned for support of other model families and custom fine-tuned models in the near future.