VCoT-Bench-Loc - Middle: leaderboard
Metric: Weighted completion accuracy (%) on the tasks missing the middle third of the verification chain of VCoT-Bench-Loc (390 tasks in the sub-benchmark), Verus proof-completion tasks built by removing verification steps from solver-lifted Rust proofs (VCoT-Lift), zero-shot greedy decoding (temperature 0 except GPT-5 mini); a completion scores 1 if Verus accepts its syntax and a GPT-5 mini judge rates it semantically equivalent to the lifted ground-truth proof, 0.5 if only semantically correct, 0.25 if only syntactically correct, 0 otherwise; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash (Preview) | 47.69 | #78 |
| 2 | Claude Sonnet 4.5 | 46.35 | #138 |
| 3 | GPT-5.2 | 41.15 | #105 |
| 4 | GPT-5 Mini (2025-08-07) | 36.92 | #165 |
| 5 | Gemini 3.1 Pro (Preview) | 36.15 | #54 |
| 6 | Claude Haiku 4.5 (20251001) | 35.19 | #251 |
| 7 | DeepSeek V3.2 | 32.31 | #198 |
| 8 | DeepSeek R1 0528 | 27.88 | #217 |
| 9 | Qwen 3 8B (Non-reasoning) | 16.54 | #667 (Qwen 3 8B) |
| 10 | Qwen 3 8B (Thinking) | 14.81 | #667 (Qwen 3 8B) |
| 11 | GPT-OSS-20B | 9.81 | #499 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=vcot-bench-loc-middle · How It Works · Data refreshed daily, snapshot 2026-10-11.