NVIDIA
NVIDIA
Language model · United States ·Apr 2025

Nemotron-H 56B

Language Open Weights
56B
Parameters
12
Benchmarks

About Nemotron-H 56B

Nemotron-H 56B is an AI model developed by NVIDIA, in the language category, released in 2025, made available as an open-weights model with 56B params.

On this page you'll find Nemotron-H 56B's full specifications, including 12 benchmark results. Review provider pricing and benchmark scores below, or compare Nemotron-H 56B head-to-head with other language models.

→ More models by NVIDIA → All language models → AI benchmarks leaderboard

Links

Announcement ↗ Hugging Face ↗

Run Nemotron-H 56B locally — what to buy

Share this build ↗

Sized to what Nemotron-H 56B actually needs (~8k context at Q4), not to the biggest GPU: Good = cheapest that runs it, Better = best value, Best = most headroom. Speed is a hardware estimate; anything past ~120 tok/s is shown as instant.

Good
Apple M1 Max 64GB
Apple · ~$2,600
~10.5 tok/s
Q4 · fits full context · up to 8k ctx
Runs with: ExLlamaV2, KoboldCpp, LM Studio
See full estimate →
Better
Apple M2 Max 96GB
Apple · ~$3,300
~10.5 tok/s
Q4 · fits full context · up to 8k ctx
Runs with: ExLlamaV2, KoboldCpp, LM Studio
See full estimate →
Best
Apple M1 Ultra 64GB
Apple · ~$3,700
~20.9 tok/s
Q4 · fits full context · up to 8k ctx
Runs with: ExLlamaV2, KoboldCpp, LM Studio
See full estimate →

Software support

·ExLlamaV2 ·KoboldCpp ·LM Studio ·MLX ·Ollama ·SGLang ·TensorRT-LLM ·llama.cpp ·vLLM ?KTransformers ?text-generation-webui

✓ measured · · compatible · — not supported. Informational only — speed is hardware-based.

Consumer GPU
6.3–7.6 tok/s
2 options →
Apple
10.5–20.9 tok/s
13 options →
Datacenter GPU
21.2–120+ tok/s
13 options →

Community

Would you run Nemotron-H 56B again?
No votes yet — be the first.
Did it run? — community reports by hardware

No reports yet. Own this model on some hardware? Be the first to confirm it runs.

Discussion

No comments yet. Start the discussion.

Nemotron-H 56B benchmark scores▶
Compare on the benchmark table →
Benchmark Category Score Variant
HumanEval code 60.37% base-0-shot
HumanEval+ code 54.27% base-0-shot
MBPP code 77.82% base-sanitized-3-shot
MBPP+ code 67.2% base-0-shot
MMLU general 84.21% base-5-shot
MMLU-Pro general 60.51% base-5-shot-cot
ARC-Challenge knowledge 94.97 base-25-shot
GSM8K math 93.71% base-8-shot-cot
MATH math 59.42% base-4-shot-cot
MATH-500 math 57.37% base-4-shot-cot
HellaSwag reasoning 89 base-10-shot
Winogrande reasoning 84.45 base-5-shot

More models like Nemotron-H 56B

AI
DMind-1
 AI
COLLECTIVE-1
 AI
Cogito 1 series
 AI
Euler (Euler大模型)
· 2.5B
 AI
SmolLM-1.7B
· 1.7B
 AI
Llama-3-70B-Synthia-v3.5
· 70B

Frequently asked questions

What is Nemotron-H 56B? ▶

Nemotron-H 56B is an AI model developed by NVIDIA, in the language category, released in 2025. It is tracked on GenAIList with its specifications, benchmark scores and provider pricing.

Who created Nemotron-H 56B? ▶

Nemotron-H 56B was developed by NVIDIA and released in 2025.

Is Nemotron-H 56B open source or proprietary? ▶

Nemotron-H 56B ships as an open-weights model: you can download and self-host the weights, though the licence may place some restrictions on use.

How much does Nemotron-H 56B cost? ▶

Pricing for Nemotron-H 56B depends on the provider. See the providers table on this page for the latest API rates.

How does Nemotron-H 56B perform on benchmarks? ▶

Nemotron-H 56B is benchmarked across 12 evaluations on GenAIList, including HumanEval (60.37%). See the full benchmark table below and compare it with other models.

No videos yet. Know a good one? Add it below.

Know a great video about Nemotron-H 56B?