VeriContest - ProofGen: leaderboard

VeriContest (2026) is a verifiable code generation benchmark of 946 LeetCode and Codeforces problems in Rust with Verus, each with expert-validated specifications, reference code and Verus-checked proofs. This task gives the specification and code and asks for the Verus proofs that make the program verify; pass@1 in percent, higher is better.

Metric: Proof generation pass@1 (%): share of problems for which the model, given the ground-truth specification and code (spec_code2proof), writes Verus proofs that verify without modifying the specification or code, on the 946 VeriContest competitive-programming problems (690 LeetCode, 256 Codeforces) in Rust with Verus, one-shot prompting with a Verus-Bench example, temperature 0 where the API exposes it (provider default for GPT-5.5 and Claude Opus 4.7); higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 10 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1GPT-5.5 (2026-04-23)13.95#39
2Gemini 3.1 Pro (Preview)13.53#99
3Claude Sonnet 4.613.21#173
4Claude Opus 4.712.68#85
5GPT-5.4 Mini (2026-03-17)7.08#288
6Gemini 3 Flash (Preview)6.34#152
7DeepSeek V4 Pro5.6#196
8GLM-4.7 Flash5.5#1349
9DeepSeek V4 Flash4.86#267
10Qwen 3.6 27B4.86#351

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=vericontest-proofgen · How It Works · Data refreshed daily, snapshot 2026-11-21.