Llama-3.3-70B-Instruct
- Name in the API
meta-llama/Llama-3.3-70B-Instruct- Base model
- meta-llama/Llama-3.3-70B-Instruct
- Weights we run
- nvidia/Llama-3.3-70B-Instruct-NVFP4 (NVFP4 quantization)
- Type
- Chat
- Input → output
- text → text
- Context length
- 65,536 tokens
- Availability
- Dedicated
- Hardware
- One NVIDIA RTX PRO 6000 Blackwell (96 GB); the FP4 weights need a Blackwell GPU
- Thinking
- No. The model answers directly.
reasoning_effort and chat_template_kwargs are accepted but have no effect.
- Tool calling
- Yes
- Example
- Chat request. Set
model to the name above.