FireWorldBench - Counterfactual Intervention Reasoning (Structured Records): leaderboard
Metric: Completion accuracy (%; P5, counterfactual intervention reasoning items of the controlled-simulation subset; structured-observation track (textual sensor traces and physical records); each cell is the mean of the matched choice-format score (Jaccard overlap of selected and gold option sets) and open-report score (exact agreement of required structured slots); deterministic evaluator, no LLM judge; failed or unparseable responses score 0). Source: arxiv.org. Saturation forecast: Around 2029. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.6 Luna | 48.07 |
| 2 | Gemma 3 27B (IT) | 46.85 |
| 3 | Qwen 3.5 Plus (2026-02-15) | 44.92 |
| 4 | DeepSeek V3.2 | 44.45 |
| 5 | Gemini 2.5 Flash | 40.41 |
| 6 | Qwen 3 32B | 39.61 |
| 7 | Kimi K2 | 38 |
| 8 | Qwen 3 235B A22B | 36.47 |
| 9 | Qwen 3 Next 80B A3B Instruct | 33.95 |
| 10 | Qwen 3 8B | 31.58 |
| 11 | Claude Haiku 4.5 | 31.45 |
| 12 | GLM 4.5 Air | 29.56 |
| 13 | Llama 3.1 8B Instruct | 29.38 |
| 14 | Gemma 3 12B (IT) | 27.89 |
| 15 | Llama 3.3 70B Instruct | 27.8 |
Interactive version: theaggregate.ai/benchmark?slug=fireworldbench-counterfactual-intervention-reasoning-structured-records · How It Works · Data refreshed daily, snapshot 2026-09-26.