Models · NVIDIA · Out sinceReleased 11 Mar 2026
Nemotron 3 Super 120B A12B
Nemotron 3 Super 120B A12B is made by NVIDIA. We don't have enough test results yet to rank it. It's cheap to use. Its makers have published it, so you can run it on your own servers.
120B total / 12B active LatentMoE (Mamba-2 + MoE + attention, MTP). NVIDIA Nemotron Open Model License (commercial use allowed). Card: up to 1M context; minimum 8x H100-80GB in BF16, NVFP4 variant runs on a single B200 / DGX Spark. Base checkpoint published. OpenRouter context 262,144 (~$0.08/$0.45).
- —
- —
- —
- Cheap$0.08 / $0.45
- $0.17
- 262K
- —
- 8 (8 independent8 indep.)
- 11 Mar 2026
- NVIDIA
- text
Scores are out of 100 for each area, compared with every model we track; “#” is its rank. Context is how much text it can read at once (1M is roughly 700,000 words). Blended price mixes the cost of what you send and what it writes back.
OpenRouter is a service that gives access to many AI models in one place. This model is listed there as nvidia/nemotron-3-super-120b-a12b.
Route it as nvidia/nemotron-3-super-120b-a12b at $0.08 in / $0.45 out per 1M tokens, 262K context. Listed since 11 Mar 2026.
frequency_penaltyinclude_reasoninglogit_biaslogprobsmax_tokensmin_ppresence_penaltyreasoningreasoning_effortrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p
Run it yourself
Self-hosting
On your own servers,Licence, sizeunder your own control.and hardware.
- Yes
- 120B / 12B
Runs on one server graphics card.
≤ 80 GB at 4-bit (one H100/H200). Weights ≈ 126 GB at 8-bit, 66 GB at 4-bit (+10–30% for KV cache). MoE: 12B active per token.
Licence conditions: Commercial use and derivatives allowed; NVIDIA claims no output ownership. On redistribution include the license and a NOTICE with 'Licensed by NVIDIA Corporation under the NVIDIA Nemotron Model License'; rights terminate on patent/copyright litigation against the model.
Restrictions: Commercial use and derivatives allowed; NVIDIA claims no output ownership. On redistribution include the license and a NOTICE with 'Licensed by NVIDIA Corporation under the NVIDIA Nemotron Model License'; rights terminate on patent/copyright litigation against the model.
Languages: Languages: Supported: English, French, German, Italian, Japanese, Spanish, Chinese; Base model metadata also lists sv, da, fi, nl, pl, cs etc.
Training it further: Fine-tuning: Megatron-LM / NeMo RL / NeMo Gym used for training; Unsloth states support for the whole Nemotron family.
Quantisations: bf16, fp8, nvfp4, gguf (community: unsloth, lmstudio-community). Engines: vLLM, SGLang.
Download (Hugging Face)Weights on Hugging Face ↗huggingface.co ↗huggingface.co ↗nvidia.com ↗
Every test result
Every result
All the numbers,with where they came from.
Every result we've found for this model, with who measured it and a link to where we read it.
All raw rows (vendor and independent kept separate), with setting label, source, date, harness and notes.
| NotesNotes | ||||||
|---|---|---|---|---|---|---|
| AA-LCR | 65.7% | DefaultNemotron 3 Super 120B A12B (Reasoning) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | AA slug nvidia-nemotron-3-super-120b-a12b |
| AA-Omniscience Index | -41.5 | DefaultNemotron 3 Super 120B A12B (Reasoning) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | AA slug nvidia-nemotron-3-super-120b-a12b; AA-Omniscience Index (-100..100) |
| Artificial Analysis Intelligence Index | 12.8 | DefaultNemotron 3 Super 120B A12B (Reasoning) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | AA slug nvidia-nemotron-3-super-120b-a12b |
| Artificial Analysis output speed | 163 tok/s | DefaultNemotron 3 Super 120B A12B (Reasoning) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | AA slug nvidia-nemotron-3-super-120b-a12b; median output tokens/sec across AA-tracked API providers (not a self-hosted measurement) |
| GPQA Diamond | 80.0% | DefaultNemotron 3 Super 120B A12B (Reasoning) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | AA slug nvidia-nemotron-3-super-120b-a12b |
| Humanity's Last Exam | 20.8% | DefaultNemotron 3 Super 120B A12B (Reasoning) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | AA slug nvidia-nemotron-3-super-120b-a12b |
| SciCode | 36.2% | DefaultNemotron 3 Super 120B A12B (Reasoning) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | AA slug nvidia-nemotron-3-super-120b-a12b |
| Terminal-Bench 2.1 | 38.6% | DefaultNemotron 3 Super 120B A12B (Reasoning) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | AA slug nvidia-nemotron-3-super-120b-a12b |