Benchmark Almanac

Beyond Survival

Gemini-2.5-pro (snapshot unspecified) 71.5% tops its leaderboard GPT-5 58.7% leads the models we track
Not in model cards

Models play social deduction games with human-aligned strategies and interaction patterns.

The 2026 EMNLP paper compares multiple language models in social deduction games. The checked page does not state the best score.

Released
November 2026
Kind
Multi-agent games
Level
Everyday
Scored
Settled by events
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Text, Actions
Interaction
Multi-agent
Environment
Werewolf
Items from
Live play or votes
Refresh
Live
Run by
Anyone
In views
Games, Agents, Kept fresh
Strongest model tested
GPT-5 #71 on the capability index
Last model added
September 2025
Scores refreshed
November 21st, 2026
15
Entries in its paper, 13 models we track

Gemini-2.5-pro (snapshot unspecified) tops the leaderboard with 71.5% GPT-5 leads with 58.7% among models released by September 2025

Best score per entry from the benchmark's paper, read on November 21st, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is GPT-5, at 58.7%. This view shows only models we track, each at its best.

The newest model with a score here came out in September 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-2.5-pro (snapshot unspecified) 71.5%
  2. 2 Deepseek-R1 (snapshot unspecified) 59%
  3. 3 GPT-5 58.7%
  4. 4 GLM-4.5 57.3%
  5. 5 DeepSeek-V3.2-Exp 56.5%
  6. 6 Gemini 2.5 Flash Sep 2025 55.8%
  7. 7 DeepSeek-V3.1 52%
  8. 8 GPT-5 mini 48.8%
  9. 9 QwQ-32B 45.6%
  10. 10 Llama 4 Scout 45%
  11. 11 Qwen3-32B 44.6%
  12. 12 Qwen3-30B-A3B 43.1%
  13. 13 gpt-oss-20b 35.8%
  14. 14 Gemma 3 27B 34.9%
  15. 15 GPT-5 nano 31.1%

Source: The Beyond Survival paper

#EntryScoreSettingDateSource
1 Gemini-2.5-pro (snapshot unspecified) 71.5% Official leaderboard
2 Deepseek-R1 (snapshot unspecified) 59% Official leaderboard
3 GPT-5 58.7% Aug 2025 Official leaderboard
4 GLM-4.5 57.3% Aug 2025 Official leaderboard
5 DeepSeek-V3.2-Exp 56.5% Sep 2025 Official leaderboard
6 Gemini 2.5 Flash Sep 2025 55.8% Sep 2025 Official leaderboard
7 DeepSeek-V3.1 52% Aug 2025 Official leaderboard
8 GPT-5 mini 48.8% Aug 2025 Official leaderboard
9 QwQ-32B 45.6% Mar 2025 Official leaderboard
10 Llama 4 Scout 45% Apr 2025 Official leaderboard
11 Qwen3-32B 44.6% Apr 2025 Official leaderboard
12 Qwen3-30B-A3B 43.1% Apr 2025 Official leaderboard
13 gpt-oss-20b 35.8% Aug 2025 Official leaderboard
14 Gemma 3 27B 34.9% Mar 2025 Official leaderboard
15 GPT-5 nano 31.1% Aug 2025 Official leaderboard
#ModelScoreSettingDateSource
1 GPT-5 58.7% Aug 2025 Official leaderboard
2 GLM-4.5 57.3% Aug 2025 Official leaderboard
3 DeepSeek-V3.2-Exp 56.5% Sep 2025 Official leaderboard
4 Gemini 2.5 Flash Sep 2025 55.8% Sep 2025 Official leaderboard
5 DeepSeek-V3.1 52% Aug 2025 Official leaderboard
6 GPT-5 mini 48.8% Aug 2025 Official leaderboard
7 QwQ-32B 45.6% Mar 2025 Official leaderboard
8 Llama 4 Scout 45% Apr 2025 Official leaderboard
9 Qwen3-32B 44.6% Apr 2025 Official leaderboard
10 Qwen3-30B-A3B 43.1% Apr 2025 Official leaderboard
11 gpt-oss-20b 35.8% Aug 2025 Official leaderboard
12 Gemma 3 27B 34.9% Mar 2025 Official leaderboard
13 GPT-5 nano 31.1% Aug 2025 Official leaderboard

Number one over time

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

OpenAI
Benchmark paper 1 leader
GPT-5 58.7%

Source: The Beyond Survival 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%) GPT-5 QwQ-32B 0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) GPT-5 QwQ-32B

Source: Official Beyond Survival 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

9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 14 months ago, in September 2025, for DeepSeek-V3.2-Exp. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
  • Pass Independent runs Scores from the benchmark’s paper (15 entries).
  • Pass Models scored 13 models we track, plus 2 other entries on its leaderboard.
  • Warning Newest score The newest came 14 months ago, in September 2025, for DeepSeek-V3.2-Exp.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5 (#71).
  • Pass Headroom The best is 58.7%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on aclanthology.org: 13 rows for models we track and 2 other entries at the last read.
  • Pass Last read November 21st, 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