# KernelBench Hard H100 (KernelBench)

> An agentic GPU-kernel benchmark that measures how much of the hardware roofline a model's correct, audit-clean kernels reach on six demanding CUDA and Triton problems.

Canonical page: https://benchlm.ai/benchmarks/kernelbench

- Category: [Coding](/coding)
- Last updated: September 23, 2026 snapshot

## About KernelBench

- Year: 2026
- Tasks: 6 GPU-kernel optimization problems
- Format: Mean peak fraction of hardware roofline over valid cells
- Difficulty: Agentic GPU systems engineering
- Paper: [KernelBench Hard](https://kernelbench.com/hard?gpu=h100)

The mirrored board comes from the independent kernelbench.com Hard suite, not the Stanford KernelBench project. One long-running agent session tackles each problem. The visible score averages peak fraction of roofline over valid cells; failed and audit-flagged cells are excluded from that mean. We keep the benchmark display only because harness, hardware, and audit state are inseparable from the score.

KernelBench is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (5 models)

| Rank | Model | Configuration | Creator | Score |
|------|-------|---------------|---------|-------|
| 1 | [Claude Fable 5](/models/claude-fable) | claude · max | Anthropic | 23.5% |
| 2 | [Claude Opus 5](/models/claude-opus-5) | or-opus · max | Anthropic | 21.8% |
| 3 | [Kimi K3](/models/kimi-k3) | kinetic-claude | Moonshot AI | 20.9% |
| 4 | [GPT-5.6 Sol](/models/gpt-5-6-sol) | codex · xhigh | OpenAI | 13.6% |
| 5 | [Qwen3.8 Max](/models/qwen3-8-max) | or-fable · xhigh | Alibaba | 13.4% |

## FAQ

### What does KernelBench measure?

An agentic GPU-kernel benchmark that measures how much of the hardware roofline a model's correct, audit-clean kernels reach on six demanding CUDA and Triton problems.

### Which model leads the published KernelBench snapshot?

Claude Fable 5 currently leads the published KernelBench snapshot with a score of 23.5%.

### How many models are evaluated on KernelBench?

The September 23, 2026 snapshot contains 5 AI models.

### Does KernelBench affect BenchLM's overall score?

Not directly. KernelBench is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.