Benchmark profile

Humanity's Last Exam without tools (HLE w/o tools)

Tool-free variant of Humanity's Last Exam that isolates a model's raw frontier reasoning.

Data verified July 23, 2026

Benchmark score on HLE w/o tools — July 23, 2026

BenchLM mirrors the published score view for HLE w/o tools. Claude Mythos 5 leads the public snapshot at 59% , followed by Muse Spark 1.1 (52.2%) and Sakana Fugu-Ultra (50%). BenchLM does not use these results to rank models overall.

1Closed

Claude Mythos 5

Anthropic

claude-mythos-5

59%

Overall 83.93Context 1M+

2Closed

Muse Spark 1.1

Benchmark score table (25 models)

Score

Claude Mythos 5Anthropic · Closed

59%

Muse Spark 1.1Meta · Closed

52.2%

Sakana Fugu-UltraSakana AI · Closed

50%

Claude Opus 4.8Anthropic · Closed

49.8%

Sakana FuguSakana AI · Closed

47.2%

Claude Opus 4.7 (Adaptive)Anthropic · Closed

46.9%

Gemini 3.1 ProGoogle · Closed

45.4%

Kimi K3Moonshot AI · Closed

43.5%

Claude Sonnet 5Anthropic · Closed

43.2%

GPT-5.5 ProOpenAI · Closed

43.1%

Muse SparkMeta · Closed

42.8%

GPT-5.4 ProOpenAI · Closed

42.7%

GPT-5.5OpenAI · Closed

41.4%

GLM-5.2Z.AI · Open weight

40.5%

Claude Opus 4.6Anthropic · Closed

40%

GPT-5.4OpenAI · Closed

39.8%

MiMo-V2.5-ProXiaomi · Closed

34%

Grok 4.20xAI · Closed

31.6%

InklingThinking Machines Lab · Open weight

30%

GPT-5.4 miniOpenAI · Closed

28.2%

Nemotron 3 UltraNVIDIA · Open weight

26.7%

GPT-5.4 nanoOpenAI · Closed

24.3%

Gemma 4 31BGoogle · Open weight

19.5%

Gemma 4 26B A4BGoogle · Open weight

8.7%

Gemma 4 12BGoogle · Open weight

5.2%

The published HLE w/o tools snapshot places Claude Mythos 5 first at 59%. The third row is 9.0 points behind. The broader top-10 range is 15.9 points, so the table still separates the published systems.

25 models have been evaluated on HLE w/o tools. The benchmark falls in the Knowledge category. This category carries a 12% weight in BenchLM.ai's overall scoring system. HLE w/o tools is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About HLE w/o tools

Year

2026

Tasks

Expert-level questions

Format

Tool-free expert QA

Difficulty

Frontier expert level

This variant removes external tools so the score reflects pure model performance on frontier expert questions.

Introducing GPT-5.4 mini and nano

BenchLM freshness & provenance

Version

HLE w/o tools 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

CurrentDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

FAQ

What does HLE w/o tools measure?

Tool-free variant of Humanity's Last Exam that isolates a model's raw frontier reasoning.

Which model scores highest on HLE w/o tools?

Claude Mythos 5 by Anthropic currently leads with a score of 59% on HLE w/o tools.

How many models are evaluated on HLE w/o tools?

25 AI models have been evaluated on HLE w/o tools on BenchLM.

Compare Top Models on HLE w/o tools

Claude Mythos 5 vs Muse Spark 1.1 Muse Spark 1.1 vs Sakana Fugu-Ultra Sakana Fugu-Ultra vs Claude Opus 4.8 Claude Opus 4.8 vs Sakana Fugu

Last updated: July 23, 2026 · BenchLM version HLE w/o tools 2026

Choose a model with this week’s evidence

Join 2,000+ readers for ranking moves, pricing changes, and the claims that still need proof.

One email each week. Unsubscribe anytime.