Nvidia's open-weight Nemotron 3.5 Lightning prioritizes speed over maximum intelligence
Nvidia's new Nemotron 3.5 Lightning is a compact open-weights model that matches OpenAI's gpt-oss-120b on intelligence benchmarks with a quarter of the parameters while delivering the fastest inference speeds in its class.
Nvidia has released Nemotron 3.5 Lightning, the first model in its new Nemotron 3.5 lineup. The model directly succeeds the Nemotron 3 Nano 30B A3B and keeps its hybrid Mamba-Transformer architecture, with 31.6 billion total parameters and only 3.6 billion active at any given time.
According to the independent benchmarking platform Artificial Analysis, the model scores 24 on the Intelligence Index, a nine-point jump from its predecessor (15). That puts Lightning on par with OpenAI's gpt-oss-120b (24) and just behind Nvidia's own Nemotron 3 Super (26), which is about four times larger. The smartest small models in the same size class, like Qwen3.6 35B A3B (32) and Meta's new Muse Glimmer (35), still hold a clear lead.
Nvidia is targeting a different spot on the efficiency frontier with Lightning. In pre-release tests using the final NVFP4 weights, the model hits nearly 670 tokens per second, the highest measured throughput among all compared models and almost twice as fast as Google's Gemini 3.5 Flash-Lite (386 tokens/s). A task from the Intelligence Index takes about 0.5 minutes to complete, while Qwen3.6 35B A3B needs around 3.5 minutes, and Gemma 4 31B takes roughly 5.8 minutes.
Proprietary models still dominate the overall efficiency frontier. Gemini 3.5 Flash-Lite scores 37 on the Intelligence Index with a similar time per task, and GPT-5.6 Luna (max) reaches 52 points in under two minutes.
Agentic benchmarks show the biggest gains
The biggest improvements show up in agentic benchmarks, according to Artificial Analysis. On GDPval-AA v2, Lightning reaches an Elo rating of 824, a 334-point gain over Nemotron 3 Nano. That beats both gpt-oss-120b (800) and the larger Nemotron 3 Super (698). On Terminal-Bench v2.1, the score jumps from 7 to 24.3 percent, nearly matching gpt-oss-120b at 26.2 percent.
Nvidia ships the model under the permissive OpenMDW-1.1 license, positioning it as a high-throughput workhorse for agent-based pipelines. Artificial Analysis reports that Nvidia worked with partners like CodeRabbit and Harvey on post-training to boost performance in specific domains.
Availability
Nvidia provides the model in both BF16 and NVFP4 weights. The NVFP4 variant also scores 24 on the Intelligence Index with minimal quality loss compared to the higher-precision version, according to Artificial Analysis. The reasoning model handles text only and supports a context window of one million tokens. Weights are available now, and serverless inference is offered by DeepInfra, Fireworks, FriendliAI, CoreWeave, GMI Cloud, Nebius, and Crusoe, among others.
Nvidia's push for efficiency over size isn't new. In a widely discussed paper last year, its researchers argued that models under 10 billion parameters can handle most agent workloads as well as 70- to 175-billion-parameter models at one-tenth to one-thirtieth the cost. Nemotron 3.5 Lightning has 31.6 billion parameters but activates only 3.6 billion per step, putting it in the same lightweight class. At nearly 670 tokens per second, it also beats gpt-oss-120b and the larger Nemotron 3 Super on agentic benchmarks, making it the clearest product-level proof of that thesis yet.
AI News Without the Hype – Curated by Humans
Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.