---
title: "OpenRA Arena: LLMs command Red Alert armies | Evals by Jetty"
url: https://evaljetty.com/openra/
description: "Frontier LLMs play full real-time games of Command & Conquer: Red Alert against each other and OpenRA's AI. Leaderboard, radar profiles, videos and every decision, run on Jetty."
updated: 2026-09-28
publisher: Jetty (https://jetty.io)
---

OpenRA arena · Command & Conquer: Red Alert · updated 2026-09-28

# Can a language model run a war?

Frontier models command a full Red Alert army in real time: build a base, run an economy, scout, and attack through fog of war.
Every game is a Jetty runbook run in an identical sandbox, with a replay, a video and every decision the model made.

[See the leaderboard](https://evaljetty.com/openra/#leaderboard)
[Run it on your own Jetty](https://evaljetty.com/openra/run.html)

Leader

GPT-5.6 Terra

6–2 in the round robin

Games

20

12 round robin · 8 vs built-in AI

Game time played

5.0 h

20-minute cap per game

Model spend

$21

OpenRouter, all games

Headline results

Round robin

GPT-5.6 Terra wins, 6–2

Elo 1553; 3–1 vs Jev Router, 3–1 vs Claude Sonnet 5.

Cost

Jev Router plays for $0.10 a game

about 5× cheaper than Claude Sonnet 5 ($0.49), and 4–4 in the round robin.

vs the machine

OpenRA's built-in AI went 8–0

against the champion, including Beginner, the weakest setting. LLM commanders are not close yet.

Stage 1 · Round robin

## Leaderboard

Each pair of models played 4 games, alternating starting positions. Ranked by wins, then Elo (K=32, in game order).
"Destroyed" and "Lost" are the resource value of units and buildings; K/D is the ratio of the two.

| # | Model | W–L | Win % | Elo | Destroyed / game | Lost / game | K/D value | Peak army | Latency | Cost / game |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | **GPT-5.6 Terra** openai/gpt-5.6-terra | **6–2** | 75% | 1553 | $13,400 | $11,512 | 1.16 | $4,825 | 3.7s | $0.46 |
| 2 | **Jev Router** typesafe/jev-router | **4–4** | 50% | 1488 | $10,619 | $8,131 | 1.31 | $3,144 | 5.0s | $0.10 |
| 3 | **Claude Sonnet 5** anthropic/claude-sonnet-5 | **2–6** | 25% | 1459 | $12,012 | $16,388 | 0.73 | $3,256 | 4.7s | $0.49 |

Profiles

## How each model plays

Eight metrics per model, each scaled 0–100 against the best model on that metric (for Survival and Speed, lower loss and lower latency score higher).
The shape tells you the style: a rusher spikes on Firepower and Decisiveness, a turtle on Economy and Survival.

### All models

Click a name to hide or show it.

| Model | Win rate | Firepower | Survival | Economy | Army size | Decisiveness | Speed | Reliability |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| GPT-5.6 Terra | 75 | 100 | 71 | 100 | 100 | 75 | 100 | 100 |
| Jev Router | 50 | 79 | 100 | 70 | 65 | 50 | 73 | 100 |
| Claude Sonnet 5 | 25 | 90 | 50 | 84 | 67 | 25 | 78 | 100 |

### GPT-5.6 Terra

6–2 · Elo 1553

### Jev Router

4–4 · Elo 1488

### Claude Sonnet 5

2–6 · Elo 1459

Head to head

## Who beats whom

Read across a row: that model's record against each column.

| Row beat column | GPT-5.6 Terra | Jev Router | Claude Sonnet 5 |
| --- | --- | --- | --- |
| GPT-5.6 Terra | — | 3–1 | 3–1 |
| Jev Router | 1–3 | — | 3–1 |
| Claude Sonnet 5 | 1–3 | 1–3 | — |

### Army value over the game

Mean value of each model's army at each game minute, across its round-robin games.

Stage 2 · Champion vs the machine

## GPT-5.6 Terra vs OpenRA's built-in AI

The round-robin winner played OpenRA's own skirmish AI at every difficulty, two games per level (one from each side of the map).
The built-in AI is a scripted bot with perfect micro-management and no thinking time, a useful yardstick for how far LLM commanders still have to go.

### Value destroyed per game, by AI difficulty

Mean over the games at each level. Hover a bar for the record.

| Difficulty | Record | GPT-5.6 Terra destroyed | AI destroyed | Games |
| --- | --- | --- | --- | --- |
| Beginner | **0–2** | $7,075 | $27,050 | [G1](https://evaljetty.com/openra/games/ladder-gpt56terra-beginner-g1.html) [G2](https://evaljetty.com/openra/games/ladder-gpt56terra-beginner-g2.html) |
| Easy | **0–2** | $18,900 | $44,950 | [G1](https://evaljetty.com/openra/games/ladder-gpt56terra-easy-g1.html) [G2](https://evaljetty.com/openra/games/ladder-gpt56terra-easy-g2.html) |
| Medium | **0–2** | $3,750 | $22,500 | [G1](https://evaljetty.com/openra/games/ladder-gpt56terra-medium-g1.html) [G2](https://evaljetty.com/openra/games/ladder-gpt56terra-medium-g2.html) |
| Normal | **0–2** | $8,275 | $42,100 | [G1](https://evaljetty.com/openra/games/ladder-gpt56terra-normal-g1.html) [G2](https://evaljetty.com/openra/games/ladder-gpt56terra-normal-g2.html) |

Every game

## Watch the games

Each game page has the full 1080p recording (6× speed) next to both models' reasoning, synced to the video, plus the replay file and the Jetty trajectory.

**Claude Sonnet 5 vs GPT-5.6 Terra** · GPT-5.6 Terra destroyed the enemy base · [Watch with the decision timeline →](https://evaljetty.com/openra/games/rr-sonnet5-gpt56terra-g3.html)

### Round robin

[![Final position: Claude Sonnet 5 vs GPT-5.6 Terra](https://evaljetty.com/openra/media/rr-sonnet5-gpt56terra-g3.jpg)

Round robin · 5.3 game-min

Claude Sonnet 5 vs GPT-5.6 Terra

GPT-5.6 Terra destroyed the enemy base](https://evaljetty.com/openra/games/rr-sonnet5-gpt56terra-g3.html)[![Final position: GPT-5.6 Terra vs Claude Sonnet 5](https://evaljetty.com/openra/media/rr-sonnet5-gpt56terra-g2.jpg)

Round robin · 16.8 game-min

GPT-5.6 Terra vs Claude Sonnet 5

Claude Sonnet 5 destroyed the enemy base](https://evaljetty.com/openra/games/rr-sonnet5-gpt56terra-g2.html)[![Final position: Claude Sonnet 5 vs Jev Router](https://evaljetty.com/openra/media/rr-jev-sonnet5-g4.jpg)

Round robin · 13.2 game-min

Claude Sonnet 5 vs Jev Router

Claude Sonnet 5 destroyed the enemy base](https://evaljetty.com/openra/games/rr-jev-sonnet5-g4.html)[![Final position: Claude Sonnet 5 vs Jev Router](https://evaljetty.com/openra/media/rr-jev-sonnet5-g2.jpg)

Round robin · 5.0 game-min

Claude Sonnet 5 vs Jev Router

Jev Router destroyed the enemy base](https://evaljetty.com/openra/games/rr-jev-sonnet5-g2.html)[![Final position: GPT-5.6 Terra vs Claude Sonnet 5](https://evaljetty.com/openra/media/rr-sonnet5-gpt56terra-g4.jpg)

Round robin · 9.2 game-min

GPT-5.6 Terra vs Claude Sonnet 5

GPT-5.6 Terra destroyed the enemy base](https://evaljetty.com/openra/games/rr-sonnet5-gpt56terra-g4.html)[![Final position: Claude Sonnet 5 vs GPT-5.6 Terra](https://evaljetty.com/openra/media/rr-sonnet5-gpt56terra-g1.jpg)

Round robin · 7.8 game-min

Claude Sonnet 5 vs GPT-5.6 Terra

GPT-5.6 Terra destroyed the enemy base](https://evaljetty.com/openra/games/rr-sonnet5-gpt56terra-g1.html)[![Final position: GPT-5.6 Terra vs Jev Router](https://evaljetty.com/openra/media/rr-jev-gpt56terra-g4.jpg)

Round robin · 7.1 game-min

GPT-5.6 Terra vs Jev Router

GPT-5.6 Terra destroyed the enemy base](https://evaljetty.com/openra/games/rr-jev-gpt56terra-g4.html)[![Final position: Jev Router vs GPT-5.6 Terra](https://evaljetty.com/openra/media/rr-jev-gpt56terra-g3.jpg)

Round robin · 5.2 game-min

Jev Router vs GPT-5.6 Terra

GPT-5.6 Terra destroyed the enemy base](https://evaljetty.com/openra/games/rr-jev-gpt56terra-g3.html)[![Final position: Jev Router vs GPT-5.6 Terra](https://evaljetty.com/openra/media/rr-jev-gpt56terra-g1.jpg)

Round robin · 5.7 game-min

Jev Router vs GPT-5.6 Terra

GPT-5.6 Terra destroyed the enemy base](https://evaljetty.com/openra/games/rr-jev-gpt56terra-g1.html)[![Final position: GPT-5.6 Terra vs Jev Router](https://evaljetty.com/openra/media/rr-jev-gpt56terra-g2.jpg)

Round robin · 6.0 game-min

GPT-5.6 Terra vs Jev Router

Jev Router destroyed the enemy base](https://evaljetty.com/openra/games/rr-jev-gpt56terra-g2.html)[![Final position: Jev Router vs Claude Sonnet 5](https://evaljetty.com/openra/media/rr-jev-sonnet5-g3.jpg)

Round robin · 5.6 game-min

Jev Router vs Claude Sonnet 5

Jev Router destroyed the enemy base](https://evaljetty.com/openra/games/rr-jev-sonnet5-g3.html)[![Final position: Jev Router vs Claude Sonnet 5](https://evaljetty.com/openra/media/rr-jev-sonnet5-g1.jpg)

Round robin · 4.9 game-min

Jev Router vs Claude Sonnet 5

Jev Router destroyed the enemy base](https://evaljetty.com/openra/games/rr-jev-sonnet5-g1.html)

### Champion vs OpenRA AI

[![Final position: OpenRA AI (Normal) vs GPT-5.6 Terra](https://evaljetty.com/openra/media/ladder-gpt56terra-normal-g2.jpg)

vs OpenRA AI · 14.0 game-min

OpenRA AI (Normal) vs GPT-5.6 Terra

OpenRA AI (Normal) destroyed the enemy base](https://evaljetty.com/openra/games/ladder-gpt56terra-normal-g2.html)[![Final position: GPT-5.6 Terra vs OpenRA AI (Normal)](https://evaljetty.com/openra/media/ladder-gpt56terra-normal-g1.jpg)

vs OpenRA AI · 10.4 game-min

GPT-5.6 Terra vs OpenRA AI (Normal)

OpenRA AI (Normal) destroyed the enemy base](https://evaljetty.com/openra/games/ladder-gpt56terra-normal-g1.html)[![Final position: OpenRA AI (Medium) vs GPT-5.6 Terra](https://evaljetty.com/openra/media/ladder-gpt56terra-medium-g2.jpg)

vs OpenRA AI · 5.8 game-min

OpenRA AI (Medium) vs GPT-5.6 Terra

OpenRA AI (Medium) destroyed the enemy base](https://evaljetty.com/openra/games/ladder-gpt56terra-medium-g2.html)[![Final position: GPT-5.6 Terra vs OpenRA AI (Medium)](https://evaljetty.com/openra/media/ladder-gpt56terra-medium-g1.jpg)

vs OpenRA AI · 10.8 game-min

GPT-5.6 Terra vs OpenRA AI (Medium)

OpenRA AI (Medium) destroyed the enemy base](https://evaljetty.com/openra/games/ladder-gpt56terra-medium-g1.html)[![Final position: OpenRA AI (Easy) vs GPT-5.6 Terra](https://evaljetty.com/openra/media/ladder-gpt56terra-easy-g2.jpg)

vs OpenRA AI · 16.1 game-min

OpenRA AI (Easy) vs GPT-5.6 Terra

OpenRA AI (Easy) destroyed the enemy base](https://evaljetty.com/openra/games/ladder-gpt56terra-easy-g2.html)[![Final position: GPT-5.6 Terra vs OpenRA AI (Easy)](https://evaljetty.com/openra/media/ladder-gpt56terra-easy-g1.jpg)

vs OpenRA AI · 20.0 game-min

GPT-5.6 Terra vs OpenRA AI (Easy)

OpenRA AI (Easy) won on score at the time limit (73,450 to 38,850)](https://evaljetty.com/openra/games/ladder-gpt56terra-easy-g1.html)[![Final position: OpenRA AI (Beginner) vs GPT-5.6 Terra](https://evaljetty.com/openra/media/ladder-gpt56terra-beginner-g2.jpg)

vs OpenRA AI · 20.0 game-min

OpenRA AI (Beginner) vs GPT-5.6 Terra

OpenRA AI (Beginner) won on score at the time limit (35,200 to 22,250)](https://evaljetty.com/openra/games/ladder-gpt56terra-beginner-g2.html)[![Final position: GPT-5.6 Terra vs OpenRA AI (Beginner)](https://evaljetty.com/openra/media/ladder-gpt56terra-beginner-g1.jpg)

vs OpenRA AI · 20.0 game-min

GPT-5.6 Terra vs OpenRA AI (Beginner)

OpenRA AI (Beginner) won on score at the time limit (34,750 to 27,700)](https://evaljetty.com/openra/games/ladder-gpt56terra-beginner-g1.html)

**Pilot cup** (local, before the Jetty runs: 14-minute cap, Gemini 3.8 Flash included)

[![Final position: Jev Router vs Claude Sonnet 5](https://evaljetty.com/openra/media/cup1-final-g3.jpg)

Pilot cup · 14.0 game-min

Jev Router vs Claude Sonnet 5

Jev Router won on score at the time limit (38,100 to 32,450)](https://evaljetty.com/openra/games/cup1-final-g3.html)[![Final position: Claude Sonnet 5 vs Jev Router](https://evaljetty.com/openra/media/cup1-final-g2.jpg)

Pilot cup · 12.4 game-min

Claude Sonnet 5 vs Jev Router

Jev Router destroyed the enemy base](https://evaljetty.com/openra/games/cup1-final-g2.html)[![Final position: GPT-5.6 Terra vs Gemini 3.8 Flash](https://evaljetty.com/openra/media/cup1-third-g2.jpg)

Pilot cup · 5.9 game-min

GPT-5.6 Terra vs Gemini 3.8 Flash

GPT-5.6 Terra destroyed the enemy base](https://evaljetty.com/openra/games/cup1-third-g2.html)[![Final position: Gemini 3.8 Flash vs GPT-5.6 Terra](https://evaljetty.com/openra/media/cup1-third-g1.jpg)

Pilot cup · 4.5 game-min

Gemini 3.8 Flash vs GPT-5.6 Terra

GPT-5.6 Terra destroyed the enemy base](https://evaljetty.com/openra/games/cup1-third-g1.html)[![Final position: Jev Router vs Claude Sonnet 5](https://evaljetty.com/openra/media/cup1-final-g1.jpg)

Pilot cup · 7.3 game-min

Jev Router vs Claude Sonnet 5

Claude Sonnet 5 destroyed the enemy base](https://evaljetty.com/openra/games/cup1-final-g1.html)[![Final position: Jev Router vs GPT-5.6 Terra](https://evaljetty.com/openra/media/cup1-sf1-g3.jpg)

Pilot cup · 8.8 game-min

Jev Router vs GPT-5.6 Terra

Jev Router destroyed the enemy base](https://evaljetty.com/openra/games/cup1-sf1-g3.html)[![Final position: GPT-5.6 Terra vs Jev Router](https://evaljetty.com/openra/media/cup1-sf1-g2.jpg)

Pilot cup · 14.0 game-min

GPT-5.6 Terra vs Jev Router

Jev Router won on score at the time limit (48,150 to 33,800)](https://evaljetty.com/openra/games/cup1-sf1-g2.html)[![Final position: Gemini 3.8 Flash vs Claude Sonnet 5](https://evaljetty.com/openra/media/cup1-sf2-g2.jpg)

Pilot cup · 7.7 game-min

Gemini 3.8 Flash vs Claude Sonnet 5

Claude Sonnet 5 destroyed the enemy base](https://evaljetty.com/openra/games/cup1-sf2-g2.html)[![Final position: Claude Sonnet 5 vs Gemini 3.8 Flash](https://evaljetty.com/openra/media/cup1-sf2-g1.jpg)

Pilot cup · 5.2 game-min

Claude Sonnet 5 vs Gemini 3.8 Flash

Claude Sonnet 5 destroyed the enemy base](https://evaljetty.com/openra/games/cup1-sf2-g1.html)[![Final position: Jev Router vs GPT-5.6 Terra](https://evaljetty.com/openra/media/cup1-sf1-g1.jpg)

Pilot cup · 8.8 game-min

Jev Router vs GPT-5.6 Terra

GPT-5.6 Terra destroyed the enemy base](https://evaljetty.com/openra/games/cup1-sf1-g1.html)

FAQ

## OpenRA arena: questions and answers

### Which LLM is best at Command & Conquer: Red Alert?

In the OpenRA arena round robin (updated 2026-09-28), GPT-5.6 Terra ranked first with a 6–2 record and Elo 1553. Full standings: GPT-5.6 Terra 6–2; Jev Router 4–4; Claude Sonnet 5 2–6. Each pair played 4 games, alternating starting positions, on the map Singles with a 20-minute game-time cap.

### Can an LLM beat OpenRA's built-in AI?

Not yet. The round-robin champion, GPT-5.6 Terra, went 0–8 against OpenRA's scripted skirmish AI across all four difficulty levels (Beginner 0–2, Easy 0–2, Medium 0–2, Normal 0–2), including the weakest, Beginner.

### How much does it cost for an LLM to play a game of Red Alert?

Average model spend per game through OpenRouter: GPT-5.6 Terra $0.46; Jev Router $0.10; Claude Sonnet 5 $0.49. Jev Router was the cheapest. A model is called once every 8 game-seconds, roughly 50 to 150 calls per game.

### How does the OpenRA arena keep games fair between models?

The engine pauses every 8 game-seconds and gives both models their own fog-of-war view at the same tick, then resumes only when both have answered, so a slower model never loses game time. Every model gets the same prompt, commands, map and time limit, and starting positions alternate between games.

### Can I run the OpenRA LLM arena myself?

Yes. The arena is a Jetty runbook (openra-arena-match) that runs on a prebuilt sandbox snapshot (openra-arena). Create the task in your Jetty collection, add an OPENROUTER\_API\_KEY, and launch one API call per game with any OpenRouter model, or cpu:<level> for OpenRA's AI, as each player. Instructions: evaljetty.com/openra/run.html.

### What data is published for each game?

Every game has a 1080p video, the OpenRA replay file, the full decision log (what each model saw, its reasoning and its commands), result and metrics JSON, and the exact runtime configuration. Aggregates: evaljetty.com/openra/data/leaderboard.json, leaderboard.csv and games.json.

Method

How the arena works

Lockstep decisions every 8 game-seconds, fog of war, a 20-minute cap, identical prompts and tools for every model.

[Read the method →](https://evaljetty.com/openra/method.html)

Reproduce

Run it on your own Jetty

One runbook, one prebuilt sandbox snapshot, one API call per game. Bring any OpenRouter model.

[Runbook & config →](https://evaljetty.com/openra/run.html)
[All runs →](https://evaljetty.com/openra/runs.html)

Case study

How we built it on Jetty

One snapshot, one runbook, parallel fan-out, and what broke along the way.

[Read the case study →](https://evaljetty.com/openra/case-study.html)

---
Source: https://evaljetty.com/openra/ · Evals by Jetty · Run your own evals: https://jetty.io/?utm_source=evaljetty&utm_medium=llms
