Nemotron 3.5 Lightning 30B A3B NVFP4
Nemotron 3.5 Lightning 30B A3B NVFP4NVIDIAOpen weightsReasoningProvisionalReleased Aug 11, 2026
Overall score
25.7
Rank 204 of 222 · 11.3–40.0
Context window
1M
1,000,000 tokens
Price per 1M tokens
— in · — out
No price listed
Blended price
—
3:1 input:output tokens
Output speed
—
Not measured yet
Category scores
Scores per capability area with the model's rank among ranked models in that category.
- 1Instruction following#7171.4
- 2Knowledge59.3
- 3Reasoning58.8
- 4Coding23.6
- 5Agentic & tool use18.4
Benchmark results (11)
As reported in system cards, papers and public evaluations. Position compares this model with every other model reporting the same benchmark.
Agentic & tool use
| Terminal-Bench 2.1 | 23.5 | #65 of 68 |
| BrowseComp | 36.8 | #49 of 50 |
| τ³-bench results | 9.5 | #21 of 23 |
| PinchBench | 83.4 | #4 of 7 |
Knowledge
| GPQA | 75.6 | #68 of 117 |
| MMLU-Pro | 81.6 | #33 of 67 |
| HLE w/o tools | 10.5 | #41 of 43 |
Coding
| SWE-bench Verified | 52.8 | #81 of 98 |
| SciCode | 31.4 | #21 of 23 |
Multilingual
| SWE Multilingual | 36.5 | #44 of 46 |
Instruction following
| IFBench | 72.9 | #27 of 42 |
Pricing by source
USD per 1M tokens.
| Source | Input | Output | Cached in |
|---|---|---|---|
| List price | — | — | — |
NVIDIA publishes this OpenMDW-1.1-licensed NVFP4 checkpoint for commercial self-hosting and does not publish a first-party hosted API token price. BenchLM represents it as self-host/free-per-token before infrastructure costs.
Speed
Measured by us through OpenRouter (median of trials).
Not available on OpenRouter, so we cannot measure it yet.
Related models
Similar score
- Phi-425.0 · $0.087
- Ministral 3 14B24.9 · $0.20
- Qwen2.5 Coder 32B Instruct26.6 · $0.74
- GPT-4o mini26.8 · $0.26
- GPT-4.1 nano24.5 · $0.17