- Services
Twenty-five hundred expert-written questions designed to be unsolvable by any current AI system, across every academic field.
HLE is the hardest broad-coverage benchmark in public use. The questions were crowdsourced from a thousand subject experts and explicitly filtered to defeat frontier models at the time of release. About 14% are multimodal, requiring image understanding. HLE measures how close a model is to the ceiling of human expert knowledge, and how much further the field still has to go.
Questions are short-answer or multiple choice. Scoring is exact-match for short-answer items and accuracy for multiple choice. Many questions include an image or diagram, so a fair score requires a multimodal model.
Benchmark data sourced from: Hugging Face, Arena.ai, Artificial Analysis.
| # | Model | Lab | Source | Score |
|---|---|---|---|---|
| 01 | Claude Opus 5.5 | Anthropic | Closed | 61.4 |
| 02 | Muse Spark Open | Meta | Open | 61.0 |
| 03 | Gemini 4 Argon | Closed | 57.1 | |
| 04 | Kimi K3 | Moonshot AI | Open | 56.0 |
| 05 | Claude Fable 5 | Anthropic | Closed | 55.5 |
| 06 | Claude Sonnet 5.5 | Anthropic | Closed | 55.0 |
| 07 | Hunyuan Hy3 | Tencent | Open | 53.2 |
| 08 | Hy3 | Tencent | Open | 53.2 |
Ad Tools for AI BuildersAll the hardware and tools you need to run private AI at your office or at home.Shop Now | ||||
| 09 | GPT-6.1 Sol | OpenAI | Closed | 52.9 |
| 10 | Kimi K3.1 Pro | MoonshotAI | Open | 52.5 |
| 11 | GPT-5.4 | OpenAI | Closed | 52.1 |
| 12 | Muse Wonder | Meta | Closed | 50.1 |
| 13 | GPT-5.6 Sol | OpenAI | Closed | 49.5 |
| 14 | Claude Opus 4.8 | Anthropic | Closed | 48.7 |
| 15 | MiniMax M3.1 Flash | MiniMax | Open | 48.0 |
The authors built it to be a benchmark that humanity might run out of room to keep designing. The questions are at or beyond the level of a top expert in each field, which makes it useful even as models improve dramatically.
Most strong open-weight models score under 10%. Frontier closed models in 2026 are between 20% and 35%. Even the best models are far from human-expert performance, which is the explicit design goal.
About 14% of items have images. Text-only models can still be evaluated on the remaining 86%, but the official score assumes full multimodal capability. Compare like with like when reading leaderboards.
93 model(s) with undisclosed parameter counts not shown. Most closed-source labs do not publish model size.
