EvoCodeBench (LeetCode) - Kotlin: leaderboard
Metric: Pass rate (%): problems of 100 whose Kotlin submission is accepted (all hidden tests within the time and memory limits); 100 recent LeetCode problems sampled across topic tags as a held-out set; a single-pass coding agent sends the problem statement and the language's starter template, extracts the code field and submits it to LeetCode's online judge, which runs the hidden tests; reasoning enabled for every model, up to 65,536 output tokens; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 7 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.5 (Thinking) | 83 | #79 (Claude Opus 4.5) |
| 2 | Gemini 3 Pro (Preview) | 82 | #64 |
| 3 | GPT-5.2 (Thinking) | 78 | #105 (GPT-5.2) |
| 4 | Gemini 3 Flash (Preview) | 75 | #78 |
| 5 | Grok 4.1 Fast (Reasoning) | 62 | #208 (Grok 4.1 Fast) |
| 6 | Claude Sonnet 4.5 (Thinking) | 42 | #138 (Claude Sonnet 4.5) |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=evocodebench-leetcode-kotlin · How It Works · Data refreshed daily, snapshot 2026-10-11.