Nemotron 4 340B Instruct
NVIDIA
Key facts · verified Oct 29, 2026
Nemotron 4 340B Instruct is a NVIDIA model released on 13 Jun 2024. Pricing: Open weights — free to self-host. Its context window is 4,096 tokens. It ranks #307 on LMArena's text leaderboard (Elo 1277, read 1 Nov 2026). Source: huggingface.co.
NVIDIA's Nemotron-4 340B Instruct, a 340-billion-parameter model published with open weights under the NVIDIA Open Model License, with a context length of 4,096 tokens.
Pricing & specs checkedSource: vendor documentation · checked Oct 29, 2026LMArena: 1277 Elo (rank 307) · source · retrieved Nov 1, 2026
Full Specifications
Open weights — free to self-host | |
|---|---|
| Identity | |
| Developer | NVIDIA |
| Released | Jun 2024 |
| Status | Open weights |
| Licence | NVIDIA Open Model License |
| Self-hostable | Yes — weights at nvidia/Nemotron-4-340B-Instruct on Hugging Face |
| Capacity | |
| Context window | 4,096 tokens |
| Capability | |
| Vision in | No — text in, text out |
| Access | |
| Fine-tuning | Yes — open weights permit it |
| Measured quality | |
| LMArena Elo | 1277 (checked Nov 2026) |
| LMArena rank | Rank 307 |