Benchmark Almanac

Group Binary Search (Riedl)

Gemini 2.0 Flash Feb 2025 75% leads since July 2025
Not in model cards

Agents each choose an integer without seeing the other guesses. They adjust after hearing whether the group sum exceeds or falls below a hidden target.

An independent Hanabi team used the plain instructions twice. Each run gave ten agents up to 50 rounds per game across 200 games.

Released
July 2025
Kind
Multi-agent games
Level
Everyday
Scored
Checked automatically
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Actions
Interaction
Multi-agent
Environment
A simulator
Refresh
Live
Run by
Anyone
Strongest model tested
Qwen3-235B-A22B-Instruct Jul 2025 #133 on the capability index
Last model added
July 2025
Scores refreshed
November 20th, 2026
4
Models scored in its paper

Gemini 2.0 Flash leads with 75% among models released by July 2025

Best score per model from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart.

The newest model with a score here came out in July 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Gemini 2.0 Flash Feb 2025, Persona 75%
  2. 2 Qwen3-235B-A22B-Instruct Jul 2025, Persona 71%
  3. 3 Llama 3.1-70B Persona + ToM 61%
  4. 4 Llama 3.1-8B Persona 14%

Source: The Group Binary Search (Riedl) paper

#ModelScoreSettingDateSource
1 Gemini 2.0 Flash Feb 2025 75% Persona Feb 2025 Official leaderboard
2 Qwen3-235B-A22B-Instruct Jul 2025 71% Persona Jul 2025 Official leaderboard
3 Llama 3.1-70B 61% Persona + ToM Jul 2024 Official leaderboard
4 Llama 3.1-8B 14% Persona Jul 2024 Official leaderboard

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
Gemini 2.0 Flash
Plain Persona + T…
75%
Qwen3-235B-A22B-Instruct
Plain Persona + T… Persona
71%
Llama 3.1-70B
Plain Persona
61%
Llama 3.1-8B
Persona + T… Persona
14%

Source: The Group Binary Search (Riedl) paper

ModelEffortHarnessScoreCost per taskOutput tokens
Gemini 2.0 Flash Feb 2025 Persona 75%
Gemini 2.0 Flash Feb 2025 Persona + ToM 71%
Qwen3-235B-A22B-Instruct Jul 2025 Persona 71%
Llama 3.1-70B Persona + ToM 61%
Gemini 2.0 Flash Feb 2025 Plain 60%
Llama 3.1-70B Persona 59%
Qwen3-235B-A22B-Instruct Jul 2025 Persona + ToM 58%
Llama 3.1-70B Plain 53%
Qwen3-235B-A22B-Instruct Jul 2025 Plain 51%
Llama 3.1-8B Persona 14%
Llama 3.1-8B Plain 11%
Llama 3.1-8B Persona + ToM 5.50%

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Google DeepMind
Benchmark paper Not tested since Jul 2025

Source: The Group Binary Search (Riedl) paper

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
In the benchmark’s paper Best so far
0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) Gemini 2.0 Flash Llama 3.1-70B 0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) Gemini 2.0 Flash Llama 3.1-70B

Source: Official Group Binary Search (Riedl) leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Games and long tasks

All 261 in Games and long tasks
  • ARC-AGI-3 In use Interactive game environments where agents must explore, infer goals and rules, and act efficiently without instructions. GPT-6.2 Astra 100%
  • Orak Fading LLM agents play 12 popular video games across most genres through an MCP interface. Each game scores in-game progress or outcomes, and the board ranks models by their average rank across the games. Gemini 2.5 Pro 3.90
  • MineDojo Fading Agents complete open-ended tasks in Minecraft, such as mining blocks and fighting mobs, from a large simulation suite. Programmatic checks score many tasks. Seed2.0 Pro 49%
  • TextQuests Only in system cards Agents play 25 classic interactive fiction games. The score is harmful actions per 100 steps, where lower is better. The board also reports game progress. Llama 4 Scout 7.00
  • GameWorld Only in system cards Multimodal game agents play 34 browser games across 170 tasks. Outcomes are checked from the game state. GPT-6.1 Terra 76.8%
  • Swarm Bench Only in system cards Models orchestrate agent swarms on complex tasks that benefit from coordinated decomposition and parallel execution. Kimi K3 76.3%

Data audit

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 16 months ago, in July 2025, for Qwen3-235B-A22B-Instruct. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the benchmark’s paper (4 entries).
  • Note Models scored Only 4 models we track.
  • Warning Newest score The newest came 16 months ago, in July 2025, for Qwen3-235B-A22B-Instruct.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Qwen3-235B-A22B-Instruct (#133).
  • Pass Headroom The best is 75%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on github.com: 12 rows for models we track at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check