Brood War Bench
Coding agents play StarCraft: Brood War against each other in real time. Systems are ranked by win rate over matches, with cost per game.
Runs agent and model pairs at several effort levels. Codex with GPT-6 Astra went 18 and 0 at the top in 2026-09.
- Level
- Expert Specialists with years of training, often over hours of work
- Scored
- Ranked by matches Models play against each other, and the results become ratings.
- Answers
- Agent task Many steps in a terminal, browser, desktop, app, game or robot.
GPT-6 Astra leads with 100% among models released by Sep 2026
Best score per entry from the benchmark's own leaderboard, read on 2 Nov 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.
The newest model with a score here came out in Sep 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- Claude Opus 5.5
- GPT-6.1 Terra
- GPT-6.1 Sol
- Claude Sonnet 5.5
- and 12 more
- 1 GPT-6 Astra xhigh 100%
- 2 Claude Fable 83.3%
- 3 GPT-5.6 Sol medium 72.2%
- 4 Claude Opus 5 66.7%
- 5 GPT-5.6 Luna low 50%
- 6 GPT-5.6 Terra xhigh 50%
- 7 Claude Sonnet 38.9%
- 8 Grok 4.6 xhigh 11.1%
- 9 Claude Haiku 0%
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-6 Astra | 100% | xhigh, Codex | Sep 2026 | Official leaderboard |
| 2 | Claude Fable | 83.3% | Official leaderboard | ||
| 3 | GPT-5.6 Sol | 72.2% | medium, Codex | Jul 2026 | Official leaderboard |
| 4 | Claude Opus 5 | 66.7% | Jul 2026 | Official leaderboard | |
| 5 | GPT-5.6 Luna | 50% | low, Codex | Jul 2026 | Official leaderboard |
| 6 | GPT-5.6 Terra | 50% | xhigh, Codex | Jul 2026 | Official leaderboard |
| 7 | Claude Sonnet | 38.9% | Official leaderboard | ||
| 8 | Grok 4.6 | 11.1% | xhigh | Aug 2026 | Official leaderboard |
| 9 | Claude Haiku | 0% | Official leaderboard |
The best score, 100%, costs $11 a task
Runs on the cost frontier score higher than any run that's cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| GPT-6 Astra | xhigh | Codex | 100% | $11 | |
| GPT-6 Astra | medium | Codex | 88.9% | $15 | |
| GPT-6 Astra | low | Codex | 77.8% | $21 | |
| GPT-5.6 Sol | medium | Codex | 72.2% | $5.12 | |
| GPT-5.6 Sol | low | Codex | 66.7% | $9.23 | |
| Claude Opus 5 | 66.7% | $21 | |||
| GPT-5.6 Sol | xhigh | Codex | 61.1% | $3.23 | |
| GPT-5.6 Luna | low | Codex | 50% | $0.42 | |
| GPT-5.6 Terra | xhigh | Codex | 50% | $2.10 | |
| GPT-5.6 Terra | medium | Codex | 44.4% | $3.15 | |
| GPT-5.6 Terra | low | Codex | 44.4% | $4.65 | |
| GPT-5.6 Luna | xhigh | Codex | 38.9% | $0.16 | |
| GPT-5.6 Luna | medium | Codex | 33.3% | $0.30 | |
| Grok 4.6 | xhigh | 11.1% | $0.66 | ||
| Grok 4.6 | medium | 5.60% | $0.79 | ||
| Grok 4.6 | low | 0% | $1.26 |
2 models have held number one since its release in Sep 2026
Lab claims count from the day the card came out, independent runs from the model’s release.
The best score passed 85% in Sep 2026
All the scores we have, one dot each, so a model tested at several settings appears more than once.
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 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 Astra 100%
- 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% - Orak Fading LLM agents play 12 popular video games across most genres through an MCP interface. Scores come from in-game progress or outcomes per title.
Nemotron 3 Nano 4B 22.9 - GameWorld Only in system cards Multimodal game agents play 34 browser games across 170 tasks. Outcomes are checked from the game state.
Gemini 3 Flash 41.9% - 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%
- TextQuests Only in system cards Agents play 25 classic interactive fiction games. Measures game progress and harmful actions per 100 steps.
Muse Spark 2.40