Enter opens the highlighted result. Esc closes.

HLE w/ tools leaderboard

As of November 18, 2026, among the results listed here, DeepSeek V4.1 Flash has the highest reported HLE w/ tools score, 63.9% (Pass@1), published by DeepSeek, ahead of Z.ai GLM 5.3 at 62.5%. 12 models have a HLE w/ tools result in our records.

Scores as the model makers published them, not re-run by us; each row links to its announcement.

Best reported score

63.9%

DeepSeek V4.1 Flash

Reported by DeepSeek, September 10, 2026 · Pass@1

Runner-up

62.5%

Z.ai GLM 5.3

Reported by Z.ai, August 18, 2026

12 models ranked · 6 makers

Reported HLE w/ tools scores

HLE w/ tools scores reported by model makers
Rank Model Score Setting Reported by Date
1 DeepSeek V4.1 Flash 63.9% Pass@1 DeepSeek September 10, 2026 (model release)
2 GLM 5.3 62.5% Not published Z.ai August 18, 2026 (model release)
3 Muse Spark Open 61.0% xhigh; w/ tools Meta October 15, 2026 (model release)
4 Step 5 Preview 59.4% High StepFun October 8, 2026 (model release)
5 Qwen3.8 Max (0902) 56.2% Not published Qwen August 3, 2026
6 GLM 5.3 Flash 55.3% Not published Z.ai August 26, 2026 (model release) · read from the maker's chart
7 GLM 5.2 54.7% w/ Tools Z.ai June 16, 2026 (model release)
8 Qwen3.7 Max 53.5% Not published Qwen August 3, 2026
9 GLM 5 50.4% w/ Tools Z.ai February 11, 2026 (model release)
10 GLM 4.7 42.8% w/ Tools Z.ai December 22, 2025 (model release)
11 MiniMax M2Retired 31.8% w/ tools MiniMax October 23, 2025 (model release)
12 GLM 4.6V 30.4% w/ Tools Z.ai December 22, 2025 (model release)

Other reported runs

HLE w/ tools scores reported by model makers, other reported runs
Model Score Setting Reported by Date
GLM 5.2 54.7% Not published Z.ai August 18, 2026 (model release)

How much to trust these numbers

Source
Reported by the model makers themselves. Not re-run by LLMBenchmark or anyone independent.
Settings
Settings differ between makers. Check the Setting column before comparing two scores.
Spread
63.9% to 30.4% across 12 models.
Freshness
Newest result published October 15, 2026; oldest October 23, 2025.
Charts
Some values were read from charts in the announcements. Two independent readings had to match exactly.

What HLE w/ tools measures

Models complete HLE tasks with tools available, and the score is the percentage of correct answers.