# LiveCodeBench — AI model rankings

Coding problems published so recently the AI can't have seen them in training — a contamination-free test of raw programming skill. Higher is better.

8 tracked models have a published LiveCodeBench score. Higher is better. Scores are as published at each model's release.

## Ranking

| Rank | Model | Developer | Score | Released |
| --- | --- | --- | --- | --- |
| 1 | DeepSeek-V4-Pro | DeepSeek | 93.5% | Apr 24 2026 |
| 2 | DeepSeek-V4-Flash | DeepSeek | 91.6% | Apr 24 2026 |
| 2 | Qwen3.7-Max | Qwen | 91.6% | May 20 2026 |
| 4 | Qwen3.8-27B | Qwen | 90.3% | Aug 14 2026 |
| 5 | Kimi K2.6 | Moonshot AI | 89.6% | Apr 21 2026 |
| 5 | Qwen3.7-Plus | Qwen | 89.6% | Jun 1 2026 |
| 7 | Kimi K2.5 | Moonshot AI | 85% | Jan 27 2026 |
| 8 | GLM-4.7 | Z.ai | 84.9% | Dec 22 2025 |


---

Canonical page: https://aireleasetracker.com/benchmark/livecodebench
Full dataset: https://aireleasetracker.com/llms-full.txt · JSON: https://aireleasetracker.com/models.json
Source: AI Release Tracker (https://aireleasetracker.com). Benchmark scores are the figures published by the releasing lab at launch.