benchmark evidence
EnigmaEval
Scale Labs puzzle-solving benchmark for long-form reasoning over public puzzle archives.
activeupstream source ->
winner on EnigmaEval
OpenAI: GPT-538.7
direct benchmark result, not a broad vertical composite | source row dated 2025-08-08
scored on 2025-12-15 · stale source data (165d)
latest mapped results | top 20
| # | Model | Score | Evidence | Tested |
|---|---|---|---|---|
| 1 | OpenAI: GPT-5 | 38.7 | model-only independent_benchmark | 2025-08-08 |
| 2 | OpenAI: GPT-5.2 | 38.4 | model-only independent_benchmark | 2025-12-15 |
| 3 | Anthropic: Claude Opus 4.5 | 17.2 | model-only independent_benchmark | 2025-11-26 |
| 4 | Anthropic: Claude Sonnet 4.5 | 12.5 | model-only independent_benchmark | 2025-10-02 |
| 5 | Anthropic: Claude Opus 4.7 | 11.9 | model-only independent_benchmark | 2025-05-23 |
| 6 | Anthropic: Claude Sonnet 4.6 | 8.1 | model-only independent_benchmark | 2025-05-23 |
| 7 | OpenAI: GPT-4.1 | 8.0 | model-only independent_benchmark | 2025-03-05 |
| 8 | Meta: Llama 4 Maverick | 2.1 | model-only independent_benchmark | 2025-04-05 |
what this result means
Scale Labs puzzle-solving benchmark for long-form reasoning over public puzzle archives.
This benchmark contributes direct public evidence. Read its scope before generalizing the result.
A win here is a win on EnigmaEval. Broad task pages require independent corroboration before naming a general winner.
source record
category: reasoning
metric: accuracy
matched models: 8
latest source date: 2025-12-15
direction: higher is better