EnigmaEval
Puzzle-hunt style multi-step reasoning. Scale AI.
As of 25 Oct 2026, Claude Fable 5 leads EnigmaEval on BenchLeader with 39.3%, ahead of GPT-5.6 Sol at 37.1%, across 5 model configurations with a published result.
- Published by
- Scale AI SEAL
- Category
- Reasoning
- Index weight
- 0.5
- Models
- 5
- Data as of
- 25 Oct 2026
- Newest result
- 25 Mar 2026
The publisher has posted no new results on this board since 25 Mar 2026. BenchLeader re-reads it every day, so newer models appear here as soon as they are published.
Scale AI SEAL Leaderboards.
What the test looks like
EnigmaEval takes puzzle-hunt puzzles, which combine wordplay, images, codes and lateral thinking over many steps, and asks for the final answer.
How it is scored
Percent of puzzles solved, published by Scale AI.
What to keep in mind
Scores are very low for every model; the puzzles are multimodal and quirky, so it is a frontier-reasoning signal rather than a general test.
- 1Claude Fable 5 (high)39.3%
- 2GPT-5.6 Sol (high)37.1%
- 3Gemini 3.1 Pro (high)36.8%
- 4Gemini 3.5 Flash (high)25.4%
- 5Claude Opus 4.8 (xhigh)23.5%
| # | ||||
|---|---|---|---|---|
| 1 |  Claude Fable 5high | 39.3% | 64.4 | 2026-03-23 |
| 2 |  GPT-5.6 Solhigh | 37.1% | 66.3 | 2026-03-25 |
| 3 |  Gemini 3.1 Prohigh | 36.8% | 60.3 | 2025-11-06 |
| 4 | Gemini 3.5 Flashhigh | 25.4% | 63.1 | 2025-11-19 |
| 5 | Claude Opus 4.8xhigh | 23.5% | – | 2026-03-10 |
More reasoning benchmarks · All LLM benchmarks
Cite as: BenchLeader, “EnigmaEval leaderboard”, data as of 25 Oct 2026, https://www.benchleader.com/benchmarks/scale_enigma_eval.
EnigmaEval: questions
- What does EnigmaEval measure?
- EnigmaEval takes puzzle-hunt puzzles, which combine wordplay, images, codes and lateral thinking over many steps, and asks for the final answer. Scores are reported in percent of tasks solved; higher is better.
- Which AI model leads EnigmaEval?
- Claude Fable 5 leads EnigmaEval with 39.3% as of 25 Oct 2026, ahead of GPT-5.6 Sol at 37.1%.
- How many models have EnigmaEval results?
- 5 model configurations have a EnigmaEval result on BenchLeader, all taken from Scale AI SEAL.
- Who runs EnigmaEval and how often is it updated?
- EnigmaEval is published by Scale AI SEAL. BenchLeader re-reads the published results once a day and records the date each result was published.
- Does EnigmaEval count toward the BenchLeader Index?
- Yes. EnigmaEval contributes to the reasoning category of the BenchLeader Index, normalised so that 50 is the average of the evaluated models.