CivBench (Civilization VI)
Agents play Civilization VI and are evaluated on strategic progress in a long-running world.
The September 2026 paper introduces a Civilization VI benchmark and compares agents. The abstract does not state the best score.
- Made by
- CivBench authors
- Released
- September 2026
- Kind
- Video games
- Level
- Expert
- Scored
- Settled by events
- Answers
- Agent task
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Agentic
- Environment
- Civilization VI, APIs and tools
- Refresh
- Live
- Run by
- Anyone
- In views
- Games, Agents, Kept fresh
- Strongest model tested
- GPT-5.4 #28 on the capability index
- Last model added
- March 2026
- Scores refreshed
- November 20th, 2026
Gemini 3.1 Pro leads with 73.0 among models released by March 2026
Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. This is the mean overall dimension score across each model’s admissible pilot games, rounded as in the dashboard. Where a model ran with several settings or harnesses, we keep its best.
The newest model with a score here came out in March 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 1
Gemini 3.1 Pro civbench_standard 73.0
- 2 Claude Opus 4.6 civbench_standard 58.0
- 3 GPT-5.4 civbench_standard 52.0
- 4 Kimi K2.5 civbench_standard 45.0
Source: Official CivBench (Civilization VI) leaderboard
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 73.0 | civbench_standard | Feb 2026 | Official leaderboard |
| 2 | Claude Opus 4.6 | 58.0 | civbench_standard | Feb 2026 | Official leaderboard |
| 3 | GPT-5.4 | 52.0 | civbench_standard | Mar 2026 | Official leaderboard |
| 4 | Kimi K2.5 | 45.0 | civbench_standard | Jan 2026 | Official leaderboard |
Rankings by part
We read 7 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Economic, Military and Scientific. Gemini 3.1 Pro leads 5 of them, and GPT-5.4 leads the other 2. Arrows show how far a model moved from its overall rank among the models we track.
Economic
4 models we track have a score here.
- 1
Gemini 3.1 Pro 21.0
- 2 Claude Opus 4.6 11.0
- 3 GPT-5.4 10.0
- 4 Kimi K2.5 7.00
Military
4 models we track have a score here.
- 1
Gemini 3.1 Pro 83.0
- 2 GPT-5.4 ↑1 79.0
- 3 Kimi K2.5 ↑1 73.0
- 4 Claude Opus 4.6 ↓2 67.0
Scientific
4 models we track have a score here.
- 1
Gemini 3.1 Pro 66.0
- 2 Claude Opus 4.6 65.0
- 3 Kimi K2.5 ↑1 63.0
- 4 GPT-5.4 ↓1 60.0
Diplomatic
4 models we track have a score here.
- 1
Gemini 3.1 Pro 81.0
- 2 GPT-5.4 ↑1 79.0
- 3 Claude Opus 4.6 ↓1 73.0
- 4 Kimi K2.5 40.0
Spatial
4 models we track have a score here.
- 1
Gemini 3.1 Pro 68.0
- 2 GPT-5.4 ↑1 60.0
- 3 Kimi K2.5 ↑1 53.0
- 4 Claude Opus 4.6 ↓2 51.0
Tool fluency
4 models we track have a score here.
- 1 GPT-5.4 ↑2 88.0
- 2 Claude Opus 4.6 87.0
- 3
Gemini 3.1 Pro ↓2 86.0
- 4 Kimi K2.5 81.0
Coherence
4 models we track have a score here.
- 1 GPT-5.4 ↑2 43.0
- 2
Gemini 3.1 Pro ↓1 25.0
- 3 Kimi K2.5 ↑1 24.0
- 4 Claude Opus 4.6 ↓2 17.0
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official CivBench (Civilization VI) leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official CivBench (Civilization VI) 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
- 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
7 of 8 checks pass, 1 warning. Look first at the Newest score check. The newest came 8 months ago, in March 2026, for GPT-5.4. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 1 pass 3 notes
- Pass Independent runs Scores from the official leaderboard (4 entries).
- Note Models scored Only 4 models we track.
- Warning Newest score The newest came 8 months ago, in March 2026, for GPT-5.4.
- Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.4 (#28).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on huggingface.co: 4 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.