VeriContest: leaderboard
Large language models can generate useful code from natural language, but their outputs come without correctness guarantees. Verifiable code generation offers a path beyond testing by requiring models to produce not only executable code, but also formal specifications and machine-checkable proofs. Progress in this direction, however, is difficult to measure: existing benchmarks are often small, fo
Metric: End-to-End (end2end) (self-reported). Source: benchmarklist.com. Saturation forecast: Around June 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 5.29 | #31 |
| 2 | Claude Sonnet 4.6 | 2.85 | #90 |
| 3 | Gemini 3.1 Pro (Preview) | 2.64 | #58 |
| 4 | Claude Opus 4.7 | 2.22 | #49 |
| 5 | DeepSeek V4 Pro | 1.06 | #102 |
| 6 | DeepSeek V4 Flash | 1.06 | #137 |
| 7 | GPT-5.4 Mini | 1.06 | #212 |
| 8 | Gemini 3 Flash | 0.95 | #99 |
| 9 | GLM-4.7 Flash | 0.21 | #509 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=vericontest · How It Works · Data refreshed daily, snapshot 2026-10-20.