OmniGameArena
Models play games while response latency advances the game clock.
The June 2026 paper compares models under latency-aware game scoring. The abstract does not state the best score.
- Made by
- OmniGameArena authors
- Released
- June 2026
- Kind
- Video games
- Level
- Everyday
- Scored
- Settled by events
- Answers
- Agent task
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Agentic
- Environment
- OmniGameArena
- Refresh
- Live
- Run by
- Anyone
- In views
- Agents, Robotics, Kept fresh
- Strongest model tested
- GPT-5.5 #21 on the capability index
- Last model added
- April 2026
- Scores refreshed
- November 20th, 2026
GPT-5.5 leads with 0.47 among models released by April 2026 GPT-5.5 leads with 0.47 among models released by April 2026
Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. The site reports each game separately. The headline is the cold-start score on ObstacleRun2D, its first solo game. 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. This view shows only models we track, each at its best.
The newest model with a score here came out in April 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 GPT-5.5 Cold start 0.47
- 2 Claude Opus 4.6 Cold start 0.33
- 3
Gemini 3 Flash Cold start 0.28
- 4 Claude Opus 4.7 Cold start 0.21
- 5 GPT-5.4 Cold start 0.12
- 6
Qwen3.5 397B-A17B Cold start 0.11
- 7 Kimi K2.5 Cold start 0.11
- 8
Gemini 3.1 Pro Cold start 0.10
- 9
Qwen3.5-122B-A10B Cold start 0.09
- 10 Claude Sonnet 4.6 Cold start 0.07
- 11 Open-P2P Cold start 0.06
- 12 NitroGen Cold start 0.03
Source: Official OmniGameArena leaderboard
- 1 GPT-5.5 Cold start 0.47
- 2 Claude Opus 4.6 Cold start 0.33
- 3
Gemini 3 Flash Cold start 0.28
- 4 Claude Opus 4.7 Cold start 0.21
- 5 GPT-5.4 Cold start 0.12
- 6
Qwen3.5 397B-A17B Cold start 0.11
- 7 Kimi K2.5 Cold start 0.11
- 8
Gemini 3.1 Pro Cold start 0.10
- 9
Qwen3.5-122B-A10B Cold start 0.09
- 10 Claude Sonnet 4.6 Cold start 0.07
Source: Official OmniGameArena leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.5 | 0.47 | Cold start | Apr 2026 | Official leaderboard |
| 2 | Claude Opus 4.6 | 0.33 | Cold start | Feb 2026 | Official leaderboard |
| 3 | | 0.28 | Cold start | Dec 2025 | Official leaderboard |
| 4 | Claude Opus 4.7 | 0.21 | Cold start | Apr 2026 | Official leaderboard |
| 5 | GPT-5.4 | 0.12 | Cold start | Mar 2026 | Official leaderboard |
| 6 | | 0.11 | Cold start | Feb 2026 | Official leaderboard |
| 7 | Kimi K2.5 | 0.11 | Cold start | Jan 2026 | Official leaderboard |
| 8 | | 0.10 | Cold start | Feb 2026 | Official leaderboard |
| 9 | | 0.09 | Cold start | Feb 2026 | Official leaderboard |
| 10 | Claude Sonnet 4.6 | 0.07 | Cold start | Feb 2026 | Official leaderboard |
| 11 | Open-P2P | 0.06 | Cold start | Official leaderboard | |
| 12 | NitroGen | 0.03 | Cold start | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.5 | 0.47 | Cold start | Apr 2026 | Official leaderboard |
| 2 | Claude Opus 4.6 | 0.33 | Cold start | Feb 2026 | Official leaderboard |
| 3 | | 0.28 | Cold start | Dec 2025 | Official leaderboard |
| 4 | Claude Opus 4.7 | 0.21 | Cold start | Apr 2026 | Official leaderboard |
| 5 | GPT-5.4 | 0.12 | Cold start | Mar 2026 | Official leaderboard |
| 6 | | 0.11 | Cold start | Feb 2026 | Official leaderboard |
| 7 | Kimi K2.5 | 0.11 | Cold start | Jan 2026 | Official leaderboard |
| 8 | | 0.10 | Cold start | Feb 2026 | Official leaderboard |
| 9 | | 0.09 | Cold start | Feb 2026 | Official leaderboard |
| 10 | Claude Sonnet 4.6 | 0.07 | Cold start | Feb 2026 | Official leaderboard |
Rankings by part
We read 24 parts from the benchmark’s own leaderboard in 3 groups: solo, coop and pvp. Each ranks models on one part of the test. GPT-5.5 leads 14 of them, and 4 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
ObstacleRun2D
10 models we track have a score here.
- 1 GPT-5.5 0.47
- 2 Claude Opus 4.6 0.33
- 3
Gemini 3 Flash 0.28
- 4 Claude Opus 4.7 0.21
- 5 GPT-5.4 0.12
- 6
Qwen3.5 397B-A17B 0.11
- 7 Kimi K2.5 0.11
- 8
Gemini 3.1 Pro 0.10
- 9
Qwen3.5-122B-A10B 0.09
- 10 Claude Sonnet 4.6 0.07
ObstacleRun3D
10 models we track have a score here.
- 1 Claude Sonnet 4.6 ↑9 0.20
- 2 Claude Opus 4.6 0.17
- 3
Gemini 3.1 Pro ↑5 0.17
- 4 GPT-5.5 ↓3 0.13
- 5
Qwen3.5 397B-A17B ↑1 0.11
- 6
Gemini 3 Flash ↓3 0.10
- 7 Claude Opus 4.7 ↓3 0.09
- 8 GPT-5.4 ↓3 0.09
- 9 Kimi K2.5 ↓2 0.07
- 10
Qwen3.5-122B-A10B ↓1 0.03
LastStand
10 models we track have a score here.
- 1 GPT-5.5 0.41
- 2 Claude Opus 4.7 ↑2 0.31
- 3 Kimi K2.5 ↑4 0.23
- 4
Gemini 3.1 Pro ↑4 0.23
- 5 GPT-5.4 0.15
- 6 Claude Opus 4.6 ↓4 0.15
- 7 Claude Sonnet 4.6 ↑3 0.14
- 8
Gemini 3 Flash ↓5 0.12
- 9
Qwen3.5 397B-A17B ↓3 0.11
- 10
Qwen3.5-122B-A10B ↓1 0.09
MonsterShoot
10 models we track have a score here.
- 1
Gemini 3.1 Pro ↑7 0.71
- 2 GPT-5.5 ↓1 0.46
- 3 Claude Opus 4.7 ↑1 0.40
- 4 Claude Opus 4.6 ↓2 0.36
- 5 GPT-5.4 0.33
- 6 Kimi K2.5 ↑1 0.29
- 7
Gemini 3 Flash ↓4 0.18
- 8 Claude Sonnet 4.6 ↑2 0.17
- 9
Qwen3.5 397B-A17B ↓3 0.07
- 10
Qwen3.5-122B-A10B ↓1 0.02
SceneEscape
10 models we track have a score here.
- 1 GPT-5.5 0.70
- 2 GPT-5.4 ↑3 0.68
- 3
Gemini 3.1 Pro ↑5 0.64
- 4 Claude Opus 4.6 ↓2 0.54
- 5 Claude Opus 4.7 ↓1 0.46
- 6 Claude Sonnet 4.6 ↑4 0.40
- 7
Gemini 3 Flash ↓4 0.28
- 8 Kimi K2.5 ↓1 0.22
- 9
Qwen3.5 397B-A17B ↓3 0.12
- 10
Qwen3.5-122B-A10B ↓1 0.06
CueChase
10 models we track have a score here.
- 1 Claude Opus 4.6 ↑1 0.84
- 2
Gemini 3.1 Pro ↑6 0.58
- 3 GPT-5.5 ↓2 0.58
- 4
Gemini 3 Flash ↓1 0.44
- 5 Claude Sonnet 4.6 ↑5 0.40
- 6 Claude Opus 4.7 ↓2 0.38
- 7 GPT-5.4 ↓2 0.30
- 8 Kimi K2.5 ↓1 0.14
- 9
Qwen3.5 397B-A17B ↓3 0.04
- 10
Qwen3.5-122B-A10B ↓1 0.04
SoloCraft
10 models we track have a score here.
- 1 GPT-5.5 0.50
- 2 Claude Opus 4.6 0.46
- 3 Claude Opus 4.7 ↑1 0.32
- 4
Gemini 3.1 Pro ↑4 0.30
- 5 Claude Sonnet 4.6 ↑5 0.25
- 6 GPT-5.4 ↓1 0.17
- 7 Kimi K2.5 0.13
- 8
Qwen3.5 397B-A17B ↓2 0.10
- 9
Gemini 3 Flash ↓6 0.07
- 10
Qwen3.5-122B-A10B ↓1 0.03
SharedFloor
10 models we track have a score here.
- 1 GPT-5.5 0.74
- 2
Gemini 3.1 Pro ↑6 0.67
- 3 Claude Opus 4.6 ↓1 0.30
- 4 Claude Sonnet 4.6 ↑6 0.30
- 5 Claude Opus 4.7 ↓1 0.27
- 6 Kimi K2.5 ↑1 0.14
- 7 GPT-5.4 ↓2 0.14
- 8
Gemini 3 Flash ↓5 0.10
- 9
Qwen3.5 397B-A17B ↓3 0.07
- 10
Qwen3.5-122B-A10B ↓1 0.02
HandoffRun
10 models we track have a score here.
- 1 GPT-5.5 0.37
- 2
Gemini 3.1 Pro ↑6 0.27
- 3 Claude Opus 4.6 ↓1 0.13
- 4 GPT-5.4 ↑1 0.12
- 5 Claude Opus 4.7 ↓1 0.08
- 6
Gemini 3 Flash ↓3 0.04
- 7 Kimi K2.5 0.02
- 8
Qwen3.5 397B-A17B ↓2 0.02
- 9 Claude Sonnet 4.6 ↑1 0.02
- 10
Qwen3.5-122B-A10B ↓1 0.00
SkyDuel vs. GPT-5.5
Win rate across ten matches. 4 models we track have a score here.
- 1 Claude Opus 4.6 ↑1 10%
- 2 Kimi K2.5 ↑5 0%
- 3
Qwen3.5 397B-A17B ↑3 0%
- 4
Gemini 3.1 Pro ↑4 0%
SkyDuel vs. Gemini 3.1 Pro Preview
Win rate across ten matches. 4 models we track have a score here.
- 1 GPT-5.5 100%
- 2 Claude Opus 4.6 90%
- 3 Kimi K2.5 ↑4 20%
- 4
Qwen3.5 397B-A17B ↑2 10%
SkyDuel vs. Kimi K2.5
Win rate across ten matches. 4 models we track have a score here.
- 1 GPT-5.5 100%
- 2 Claude Opus 4.6 80%
- 3
Gemini 3.1 Pro ↑5 80%
- 4
Qwen3.5 397B-A17B ↑2 30%
SkyDuel vs. Qwen3.5-397B-A17B
Win rate across ten matches. 4 models we track have a score here.
- 1 GPT-5.5 100%
- 2 Claude Opus 4.6 90%
- 3
Gemini 3.1 Pro ↑5 90%
- 4 Kimi K2.5 ↑3 70%
SkyDuel vs. Claude Opus 4.6
Win rate across ten matches. 4 models we track have a score here.
- 1 GPT-5.5 90%
- 2 Kimi K2.5 ↑5 20%
- 3
Qwen3.5 397B-A17B ↑3 10%
- 4
Gemini 3.1 Pro ↑4 10%
CrystalGuard vs. GPT-5.5
Win rate across ten matches. 4 models we track have a score here.
- 1 Kimi K2.5 ↑6 0%
- 2 Claude Opus 4.6 0%
- 3
Qwen3.5 397B-A17B ↑3 0%
- 4
Gemini 3.1 Pro ↑4 0%
CrystalGuard vs. Gemini 3.1 Pro Preview
Win rate across ten matches. 4 models we track have a score here.
- 1 GPT-5.5 100%
- 2 Claude Opus 4.6 60%
- 3 Kimi K2.5 ↑4 30%
- 4
Qwen3.5 397B-A17B ↑2 0%
CrystalGuard vs. Kimi K2.5
Win rate across ten matches. 4 models we track have a score here.
- 1 Claude Opus 4.6 ↑1 100%
- 2 GPT-5.5 ↓1 100%
- 3
Gemini 3.1 Pro ↑5 70%
- 4
Qwen3.5 397B-A17B ↑2 30%
CrystalGuard vs. Qwen3.5-397B-A17B
Win rate across ten matches. 4 models we track have a score here.
- 1
Gemini 3.1 Pro ↑7 100%
- 2 GPT-5.5 ↓1 100%
- 3 Claude Opus 4.6 ↓1 80%
- 4 Kimi K2.5 ↑3 70%
CrystalGuard vs. Claude Opus 4.6
Win rate across ten matches. 4 models we track have a score here.
- 1 GPT-5.5 100%
- 2
Gemini 3.1 Pro ↑6 40%
- 3
Qwen3.5 397B-A17B ↑3 20%
- 4 Kimi K2.5 ↑3 0%
MidlineClash vs. GPT-5.5
Win rate across ten matches. 4 models we track have a score here.
- 1
Gemini 3.1 Pro ↑7 50%
- 2 Kimi K2.5 ↑5 10%
- 3 Claude Opus 4.6 ↓1 0%
- 4
Qwen3.5 397B-A17B ↑2 0%
MidlineClash vs. Gemini 3.1 Pro Preview
Win rate across ten matches. 4 models we track have a score here.
- 1 GPT-5.5 50%
- 2 Claude Opus 4.6 10%
- 3 Kimi K2.5 ↑4 0%
- 4
Qwen3.5 397B-A17B ↑2 0%
MidlineClash vs. Kimi K2.5
Win rate across ten matches. 4 models we track have a score here.
- 1
Gemini 3.1 Pro ↑7 100%
- 2 Claude Opus 4.6 90%
- 3 GPT-5.5 ↓2 90%
- 4
Qwen3.5 397B-A17B ↑2 20%
MidlineClash vs. Qwen3.5-397B-A17B
Win rate across ten matches. 4 models we track have a score here.
- 1 Claude Opus 4.6 ↑1 100%
- 2
Gemini 3.1 Pro ↑6 100%
- 3 GPT-5.5 ↓2 100%
- 4 Kimi K2.5 ↑3 80%
MidlineClash vs. Claude Opus 4.6
Win rate across ten matches. 4 models we track have a score here.
- 1 GPT-5.5 100%
- 2
Gemini 3.1 Pro ↑6 90%
- 3 Kimi K2.5 ↑4 10%
- 4
Qwen3.5 397B-A17B ↑2 0%
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official OmniGameArena 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 OmniGameArena 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
8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 7 months ago, in April 2026, for GPT-5.5. 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 official leaderboard (12 entries).
- Pass Models scored 10 models we track, plus 2 other entries on its leaderboard.
- Warning Newest score The newest came 7 months ago, in April 2026, for GPT-5.5.
- Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.5 (#21).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on mxlin043.github.io: 10 rows for models we track and 2 other entries 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.