Skip to main content
Radar

Five or fewer confirmed AI changes, with original sources, on mornings when something changed.A free source-linked morning brief.

Start free brief
BenchLM
Advertise

Benchmark profile

KernelBench Internal

xAI's internal implementation of a GPU-kernel generation evaluation.

Data verified 26 confirmed releases in the last 30 daysStart free brief

Benchmark score on KernelBench Internal — August 12, 2026

BenchLM mirrors the published score view for KernelBench Internal. GPT-5.6 Sol leads the public snapshot at 73.1% , followed by Grok 4.6 (70.6%) and Kimi K3 (68.5%). We do not use these results to rank models overall.

1Closed

GPT-5.6 Sol

OpenAI

gpt-5-6-sol

73.1%
Overall 81.23Context 1.05M
2Closed

Grok 4.6

xAI

grok-4-6

70.6%
Overall 62.44Context 500K
3Closed

Kimi K3

Moonshot AI

kimi-3

68.5%
Overall 80.14Context 1.05M
5 modelsCodingCurrentDisplay onlyUpdated August 12, 2026

Benchmark score table (5 models)

Score
1
GPT-5.6 SolOpenAI · Closed
73.1%
2
Grok 4.6xAI · Closed
70.6%
3
Kimi K3Moonshot AI · Closed
68.5%
4
Claude Opus 4.8Anthropic · Closed
67.6%
5
Grok 4.5xAI · Closed
58.4%

The published KernelBench Internal snapshot places GPT-5.6 Sol first at 73.1%. The third row is 4.6 points behind. The broader top-10 range is 14.7 points, so the table still separates the published systems.

5 models have been evaluated on KernelBench Internal. The benchmark falls in the Coding category. This category carries a 20% weight in BenchLM.ai's overall scoring system. KernelBench Internal is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About KernelBench Internal

Year

2026

Tasks

GPU-kernel generation tasks

Format

Task score

Difficulty

Expert performance engineering

This lane is separate from public KernelBench fields because xAI labels the implementation internal. The comparison remains display-only.

Grok 4.6 Model Card

BenchLM freshness & provenance

Version

KernelBench Internal 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

CurrentDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

FAQ

What does KernelBench Internal measure?

xAI's internal implementation of a GPU-kernel generation evaluation.

Which model scores highest on KernelBench Internal?

GPT-5.6 Sol by OpenAI currently leads with a score of 73.1% on KernelBench Internal.

How many models are evaluated on KernelBench Internal?

5 AI models have been evaluated on KernelBench Internal on BenchLM.

Compare Top Models on KernelBench Internal

GPT-5.6 Sol vs Grok 4.6Grok 4.6 vs Kimi K3Kimi K3 vs Claude Opus 4.8Claude Opus 4.8 vs Grok 4.5
Last updated: August 12, 2026 · BenchLM version KernelBench Internal 2026

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 2,000+ readers.

One email each week. Unsubscribe anytime.