# Humanity's Last Exam (with tools) — AI model rankings

Humanity's Last Exam — extremely hard expert questions across many subjects. “With tools” means the AI is allowed to search the web or run code while answering. Higher is better.

54 tracked models have a published Humanity's Last Exam (with tools) score. Higher is better. Scores come from published lab reports and benchmark sources; recorded sources appear beside the scores.

## Ranking

| Rank | Model | Developer | Score | Source | Released |
| --- | --- | --- | --- | --- | --- |
| 1 | Kimi K3.1 Pro | Moonshot AI | 70% | Lab | Oct 23 2026 |
| 2 | Claude Opus 5.5 | Anthropic | 67.7% | Lab | Sep 22 2026 |
| 3 | Claude Fable 5.1 | Anthropic | 65.6% | Lab | Sep 1 2026 |
| 4 | Claude Fable 5 | Anthropic | 64.5% | [BenchLM](https://benchlm.ai), retrieved 2026-07-11 | Jun 9 2026 |
| 4 | Claude Mythos 5 | Anthropic | 64.5% | [BenchLM](https://benchlm.ai), retrieved 2026-08-24 | Jun 9 2026 |
| 4 | Claude Sonnet 5.5 | Anthropic | 64.5% | Lab | Sep 28 2026 |
| 7 | DeepSeek-V4.1-Flash | DeepSeek | 63.9% | Lab | Sep 10 2026 |
| 8 | Claude Opus 5 | Anthropic | 63.6% | Lab | Jul 24 2026 |
| 9 | GLM-5.3 | Z.ai | 62.5% | Lab | Aug 14 2026 |
| 10 | Muse Spark 1.1 | Meta | 62.1% | Lab | Jul 9 2026 |
| 11 | DeepSeek-V4-Pro-0813 | DeepSeek | 60% | Lab | Aug 13 2026 |
| 12 | GPT-5.4-Pro | OpenAI | 58.7% | [BenchLM](https://benchlm.ai), retrieved 2026-07-11 | Mar 5 2026 |
| 13 | Claude Opus 4.8 | Anthropic | 57.9% | Lab | May 28 2026 |
| 14 | Claude Sonnet 5 | Anthropic | 57.4% | Lab | Jun 30 2026 |
| 14 | Claude Haiku 5.5 | Anthropic | 57.4% | Lab | Oct 7 2026 |
| 16 | GPT-5.5-Pro | OpenAI | 57.2% | [BenchLM](https://benchlm.ai), retrieved 2026-07-11 | Apr 23 2026 |
| 17 | Kimi K3 | Moonshot AI | 56% | Lab | Jul 16 2026 |
| 18 | GLM-5.3-Flash | Z.ai | 55.3% | Lab | Aug 26 2026 |
| 19 | Claude Opus 4.7 | Anthropic | 54.7% | Lab | Apr 16 2026 |
| 19 | GLM-5.2 | Z.ai | 54.7% | [BenchLM](https://benchlm.ai), retrieved 2026-07-11 | Jun 16 2026 |
| 21 | Claude Opus 4.6 | Anthropic | 53% | [BenchLM](https://benchlm.ai), retrieved 2026-07-11 | Feb 5 2026 |
| 22 | GLM-5.1 | Z.ai | 52.3% | [BenchLM](https://benchlm.ai), retrieved 2026-07-11 | Apr 7 2026 |
| 23 | GPT-5.5 | OpenAI | 52.2% | Lab | Apr 23 2026 |
| 24 | GPT-5.4 | OpenAI | 52.1% | [BenchLM](https://benchlm.ai), retrieved 2026-08-31 | Mar 5 2026 |
| 25 | Gemini 3.1 Pro | Google | 51.4% | Lab | Feb 19 2026 |
| 26 | GLM-5 | Z.ai | 50.4% | Lab | Feb 11 2026 |
| 26 | Muse Spark | Meta | 50.4% | Lab | Apr 8 2026 |
| 28 | Claude Sonnet 4.6 | Anthropic | 49% | [BenchLM](https://benchlm.ai), retrieved 2026-08-31 | Feb 17 2026 |
| 29 | Qwen3.8-Max | Qwen | 43.6% | [BenchLM](https://benchlm.ai), retrieved 2026-08-31 | Aug 3 2026 |
| 30 | GLM-4.7 | Z.ai | 42.8% | Lab | Dec 22 2025 |
| 31 | DeepSeek-V4.1-Flash-1017 | DeepSeek | 42.6% | [BenchLM](https://benchlm.ai), retrieved 2026-10-19 | Oct 17 2026 |
| 32 | GPT-5.4 mini | OpenAI | 41.5% | [BenchLM](https://benchlm.ai), retrieved 2026-08-31 | Mar 17 2026 |
| 33 | Qwen3.7-Max | Qwen | 41.4% | [BenchLM](https://benchlm.ai), retrieved 2026-08-31 | May 20 2026 |
| 34 | Gemini 3.5 Flash | Google | 40.2% | [BenchLM](https://benchlm.ai), retrieved 2026-08-31 | May 19 2026 |
| 35 | GPT-5.4 nano | OpenAI | 37.7% | [BenchLM](https://benchlm.ai), retrieved 2026-08-31 | Mar 17 2026 |
| 36 | Qwen3.8-Flash-Next | Qwen | 35.9% | [BenchLM](https://benchlm.ai), retrieved 2026-08-31 | Aug 26 2026 |
| 37 | Grok 4.3 Beta | SpaceXSI | 35% | [BenchLM](https://benchlm.ai), retrieved 2026-09-22 | Apr 17 2026 |
| 38 | DeepSeek-V4-Flash-0731 | DeepSeek | 34.8% | [BenchLM](https://benchlm.ai), retrieved 2026-08-31 | Jul 31 2026 |
| 39 | Kimi K2.6 | Moonshot AI | 34.7% | [BenchLM](https://benchlm.ai), retrieved 2026-08-31 | Apr 20 2026 |
| 39 | Qwen3.7-Plus | Qwen | 34.7% | [BenchLM](https://benchlm.ai), retrieved 2026-08-31 | Jun 1 2026 |
| 41 | Claude Opus 4.5 | Anthropic | 30.8% | [BenchLM](https://benchlm.ai), retrieved 2026-08-31 | Nov 24 2025 |
| 41 | Qwen3.8-27B | Qwen | 30.8% | [BenchLM](https://benchlm.ai), retrieved 2026-08-31 | Aug 14 2026 |
| 43 | Kimi K2.5 | Moonshot AI | 30.1% | [BenchLM](https://benchlm.ai), retrieved 2026-08-31 | Jan 27 2026 |
| 44 | Nemotron 3.5 Super VL | NVIDIA | 29.2% | [BenchLM](https://benchlm.ai), retrieved 2026-10-22 | Oct 15 2026 |
| 45 | Qwen3.6-Plus | Qwen | 28.8% | [BenchLM](https://benchlm.ai), retrieved 2026-08-31 | Apr 2 2026 |
| 46 | Nemotron 3 Ultra | NVIDIA | 26.7% | [BenchLM](https://benchlm.ai), retrieved 2026-08-31 | Jun 4 2026 |
| 47 | Gemma 4 | Google | 26.5% | [BenchLM](https://benchlm.ai), retrieved 2026-10-22 | Apr 2 2026 |
| 48 | gpt-oss-120b | OpenAI | 19% | Lab | Aug 5 2025 |
| 49 | Gemini 2.5 Pro | Google | 18.8% | [BenchLM](https://benchlm.ai), retrieved 2026-08-31 | Mar 25 2025 |
| 50 | DeepSeek R1-0528 | DeepSeek | 17.7% | [BenchLM](https://benchlm.ai), retrieved 2026-10-19 | May 28 2025 |
| 51 | gpt-oss-20b | OpenAI | 17.3% | Lab | Aug 5 2025 |
| 52 | DeepSeek V3.1 | DeepSeek | 15.9% | [BenchLM](https://benchlm.ai), retrieved 2026-10-12 | Aug 21 2025 |
| 53 | Nemotron 3 Nano | NVIDIA | 10.6% | [BenchLM](https://benchlm.ai), retrieved 2026-10-22 | Dec 15 2025 |
| 54 | Kimi K2 | Moonshot AI | 4.7% | [BenchLM](https://benchlm.ai), retrieved 2026-10-12 | Jul 11 2025 |


---

Canonical page: https://aireleasetracker.com/benchmark/hle-with-tools
Site index: https://aireleasetracker.com/llms.txt
Source: AI Release Tracker (https://aireleasetracker.com). Most benchmark scores come from lab launch material; gathered results identify the leaderboard that published them.