Human Behavior Bench
Gemini 4 Argon 80.6 leads since Oct 2026 Not in model cards Models predict how specified groups of U.S. adults answer closed survey questions, as a distribution over options. Scored as 1 minus total variation distance.
1,199 tasks, 239 public and 960 private, from surveys run June to September 2026. Launched 2026-10-20.
Made by Rehearsals with Google DeepMind and Kaggle Released Oct 2026 Real-world events Public tasks plus a private test set
Last model added Sep 2026
Scores refreshed 2 Nov 2026
- Level
- Expert Specialists with years of training, often over hours of work
- Scored
- Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
- Answers
- Prediction Probabilities for events that have not happened yet.
11
Models scored on its own leaderboard
Gemini 4 Argon leads with 80.6
Best score per model from the benchmark's own leaderboard, read on 2 Nov 2026. Where a model ran with several settings or harnesses, we keep its best.
- 1 Gemini 4 Argon 80.6
- 2 Claude Opus 5 80.2
- 3 Gemini 3.7 Flash 79.7
- 4 GPT-5.6 Sol 79.5
- 5 Claude Opus 5.5 79.3
- 6 Gemini 3.8 Flash 79.2
- 7 Gemini 3 Flash 78.5
- 8 GPT-5.6 Terra 77.8
- 9 Gemini 3.5 Flash-Lite 75.9
- 10 Grok 4.7 75.2
- 11 Grok 4.20 non-reasoning 73.5
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Gemini 4 Argon | 80.6 | Sep 2026 | Official leaderboard | |
| 2 | Claude Opus 5 | 80.2 | Jul 2026 | Official leaderboard | |
| 3 | Gemini 3.7 Flash | 79.7 | Aug 2026 | Official leaderboard | |
| 4 | GPT-5.6 Sol | 79.5 | Jul 2026 | Official leaderboard | |
| 5 | Claude Opus 5.5 | 79.3 | Sep 2026 | Official leaderboard | |
| 6 | Gemini 3.8 Flash | 79.2 | Sep 2026 | Official leaderboard | |
| 7 | Gemini 3 Flash | 78.5 | Dec 2025 | Official leaderboard | |
| 8 | GPT-5.6 Terra | 77.8 | Jul 2026 | Official leaderboard | |
| 9 | Gemini 3.5 Flash-Lite | 75.9 | Jul 2026 | Official leaderboard | |
| 10 | Grok 4.7 | 75.2 | Sep 2026 | Official leaderboard | |
| 11 | Grok 4.20 | 73.5 | non-reasoning | Feb 2026 | Official leaderboard |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Google DeepMind
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
On the benchmark’s own leaderboard
No lab model card reports it yet
We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark.
More in Forecasting
- ForecastBench In use Models give probabilities for 500 auto-generated questions per round about future events, drawn from prediction markets and time-series datasets. Forecasts are scored with difficulty-adjusted Brier scores once the questions resolve. Claude 3 Haiku 44.7
- Prophet Arena In use Models and agents forecast open Kalshi prediction-market events and are scored once the events resolve. The live board ranks a difficulty-adjusted Arena score and also reports calibration error and market return. Claude Opus 5.5 86.0
- Echo General AI Prediction Leaderboard Not in model cards Models forecast unresolved events drawn from prediction markets, synthesized news questions and expert-written questions. Forecasts made at aligned times are compared in pairwise battles, and an Elo fit ranks the models.
DeepSeek-V4-Pro 1,018 - BTF-3 Not in model cards Agents forecast already-resolved binary and numeric questions while researching a frozen snapshot of the web. Scoring pools Brier and ranked probability scores. Claude Opus 5 0.12
- BackSearch Not in model cards Paid API that searches a frozen web archive as of a given date, for forecasting backtests and reproducible web-enabled evals. A tool, not a benchmark.
- BallotBench Not in model cards Nine frontier agents forecast 863 contests in the 2026 US midterms each day until polls close. Proper scoring rules grade forecasts against certified results, beside Kalshi, Polymarket and expert ratings.