whatmodel.sh / Nemotron-4 340B Instruct
Nemotron-4 340B Instruct
↗ share
+ contribute
Nemotron-4 340B Instruct is an open-weights language model from NVIDIA, released June 2024. Its published size is 340B parameters. Licence: nvidia-open-model-license.
Catalogue identity notes
Comparison-only. Arena entry 'nemotron-4-340b-instruct' (text). LMArena's leaderboard table (arena-leaderboard space @6f2eeeb) labels 'nemotron-4-340b-instruct' as 'Nemotron-4-340B-Instruct' (Nvidia, NVIDIA Open Model), linking https://huggingface.co/nvidia/Nemotron-4-340B-Instruct. Maker repo nvidia/Nemotron-4-340B-Instruct @ac75bfb: licence tag other (nvidia-open-model-license). Parameters: card badge 'Params-340B'. Released 2024-06: month the maker's Hugging Face repo was created (createdAt 2024-06-13).
$ how it compares
The LLMs ranked either side of Nemotron-4 340B Instruct, out of 828 with a score, then more from NVIDIA.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 381 | Llama 3.1 Nemotron 70B Instruct | 52±1 | local | ||
| 382 |  Palmyra X4 (palmyra-x-004) | 52±4 | comparison only | ||
| 383 | Nemotron-4 340B Instruct | 52±2 | comparison only | ||
| 384 |  Aya Expanse 32B | 52±2 | api $0.75/M +local | ||
| 385 |  Gemma 3 12B | 52±1 | api $0.075/M +local | ||
| more from NVIDIA | |||||
| 98 | Nemotron 3 Ultra 550B-A55B | 69±1 | api $1.35/M +local | ||
| 155 | Nemotron Cascade 2 30B-A3B | 66±1 | local | ||
| 168 | Nemotron 3 Super 120B-A12B | 65±1 | api $0.28/M +local | ||
$ what it's good at
chat & writing▁▁▁▃▅▄▆█▇▇▇▃ #292 of 463 * · 59% of its score · 2 tests
Beats Mistral Small 4 (119B-A6.5B), Ministral 3 14B Reasoning 2512, Qwen2.5 Coder 32B Instruct here, though they rank higher overall. Trails Flan-T5 XXL, Granite 4.0 H 1B, Granite 4.0 1B here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Polyrating rating (bias-aware re-fit of Chatbot Arena votes, mid-2024) | 1222.0 ✎ | ▂▁▂▁▃▇█▆▇▄▃▂ beats 88/95 | 30% |
| LMArena (text) | 1277.1 ✎ | ▁▁▁▂▃▃▃▆▆▇█▃ beats 110/379 | 29% |
other tests2 not in a task · 40% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| WildBench v2 WB-Score (task macro) | 47.6725 WB-Score ✎ | ▁·▁▁·▃█▄▇▆▅▆ beats 37/52 | 33% |
| Open PT-LLM Leaderboard average (9 Portuguese tasks) | 77.8% ✎ | ▃▁▁▁▁▁▁▁▂█▅▁ beats 241/276 | 6.2% |
$ where to run it
No verified API or download recorded yet · add one
Official weights and documentation: huggingface.co/nvidia ↗
$ how the score was worked out · WildBench v2 leaderboard data… 33%, Polyrating: A Cost-Effective… 30%, LMArena leaderboard dataset 29% and 1 more
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 WildBench v2 leaderboard data (Ai2) | 1 | 33% |
 Polyrating: A Cost-Effective and Bias-Aware Rating System for LLM Evaluation (arXiv:2409.00696v3) | 1 | 30% |
 LMArena leaderboard dataset | 1 | 29% |
 Open Portuguese LLM Leaderboard (archived) | 1 | 6% |
| prior · pull toward the middle | 1% | |
| weighted average | = 51.7 |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json