Translate
Independent model intelligence

Find the right AI model,
not just the loudest one.

Compare current model capability, context, modalities, public preference and API economics in one continuously refreshed decision surface.

LiveBench 2.0 LMArena models.dev OpenRouter AIHOT cross-check
Back to leaderboard
NVIDIA · Active

Nemotron Lightning 3.5

nemotron family · launched 2026-08-15

ReasoningToolsStructured
AITopic score29.8Not yet ranked

Nemotron-Lightning-3.5-30B-A3B is a 30B-parameter Mixture-of-Experts language model (3B active) from NVIDIA's Nemotron-H family, built on a hybrid Mamba-Transformer architecture for efficient long-context inference. Like other models in the family, it responds to queries by first generating a reasoning trace and then concluding with a final response, with reasoning behavior configurable through a flag in the chat template. It includes a multi-token prediction (MTP) speculative decoding head for low-latency serving.

Evidence coverage90%4 evaluations · 1 institutions
Context / maximum output262.1k / 262.1ktokens
Cache / input / output$0.01 · $0.05 · $0.2USD per million tokens
Value / freshness49.4 / 83relative value · release freshness
Product profile

What this model supports

Inputtext
Outputtext
Knowledge cutoffNot disclosed
WeightsProprietary / hosted
Capability profile

Strengths by domain

Coding22.61 public results
Research & reasoning42.91 public results
Professional work23.81 public results
Knowledge & instruction40.01 public results
Behind the score

Every result has a source

Raw results and standardized readings are shown together. Missing tests add neither points nor penalties.

Coding1 results

LiveBench 2.0 · CodingLiveBench22.60 scoreReading 22.6

Imported from LiveBench 2.0. The published 0–100 score is used directly; it is not z-score normalized.

Configuration: LiveBench 2.0Open evidence

Research & reasoning1 results

LiveBench 2.0 · ResearchLiveBench42.90 scoreReading 42.9

Imported from LiveBench 2.0. The published 0–100 score is used directly; it is not z-score normalized.

Configuration: LiveBench 2.0Open evidence

Professional work1 results

LiveBench 2.0 · ProfessionalLiveBench23.83 scoreReading 23.8

Imported from LiveBench 2.0. The published 0–100 score is used directly; it is not z-score normalized. Professional = long-agentic, browser, data-analysis and design average.

Configuration: LiveBench 2.0Open evidence

Knowledge & instruction1 results

LiveBench 2.0 · KnowledgeLiveBench40.00 scoreReading 40.0

Imported from LiveBench 2.0. The published 0–100 score is used directly; it is not z-score normalized. Knowledge uses the published LiveBench research suite.

Configuration: LiveBench 2.0Open evidence

Human preference & indices2 results

LiveBenchLiveBench28.69 scoreReading 28.7

Imported from LiveBench 2.0. The published 0–100 score is used directly; it is not z-score normalized.

Configuration: LiveBench 2.0Open evidence
AA IndexArtificial Analysis28.69 indexReading 27.7

Imported from Artificial Analysis. The published 0–100 score is used directly; it is not z-score normalized.

Configuration: LiveBench 2.0Open evidence
Same builder

More from NVIDIA