Real-Time Reasoning Gym
Models act in games while token generation consumes part of the game clock.
The November 2025 paper compares models under a real-time token budget. The abstract does not state the best score.
- Made by
- Real-Time Reasoning Gym authors
- Released
- November 2025
- Kind
- Video games
- Level
- Everyday
- Scored
- Settled by events
- Answers
- Agent task
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Agentic
- Environment
- A simulator
- Items from
- Generated
- Refresh
- One fixed set
- Run by
- Anyone
- Strongest model tested
-
DeepSeek-V3.2 #91 on the capability index
- Last model added
- December 2025
- Scores refreshed
- November 20th, 2026
DeepSeek-R1 (snapshot unspecified) tops the leaderboard with 0.92 DeepSeek-V3.2 leads with 0.81
Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. The headline averages normalized reward across Freeway, Snake, and Overcooked. 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 best model we track is DeepSeek-V3.2, at 0.81. This view shows only models we track, each at its best.
- 1 DeepSeek-R1 (snapshot unspecified) Medium, 32k tokens per step (budget sweep) 0.92
- 2 DeepSeek-V3 + DeepSeek-R1 Medium, 32k tokens per step (budget sweep) 0.89
- 3
DeepSeek-V3.2 Hard, 32k tokens per step (budget sweep) 0.81
Source: The Real-Time Reasoning Gym paper
Source: The Real-Time Reasoning Gym paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | DeepSeek-R1 (snapshot unspecified) | 0.92 | Medium, 32k tokens per step (budget sweep), Planning (R1) | Official leaderboard | |
| 2 | DeepSeek-V3 + DeepSeek-R1 | 0.89 | Medium, 32k tokens per step (budget sweep), AgileThinker | Official leaderboard | |
| 3 | | 0.81 | Hard, 32k tokens per step (budget sweep), Planning (thinking on) | Dec 2025 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 0.81 | Hard, 32k tokens per step (budget sweep), Planning (thinking on) | Dec 2025 | Official leaderboard |
Scores at each effort level
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: The Real-Time Reasoning Gym paper
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| Hard, 32k tokens per step (budget sweep) | Planning (thinking on) | 0.81 | |||
| Hard, 32k tokens per step (budget sweep) | AgileThinker | 0.77 | |||
| Hard, 16k tokens per step (budget sweep) | AgileThinker | 0.74 | |||
| Hard, 8k tokens per step (budget sweep) | AgileThinker | 0.55 | |||
| Hard, 16k tokens per step (budget sweep) | Planning (thinking on) | 0.44 | |||
| Hard, 4k tokens per step (budget sweep) | Reactive (thinking off) | 0.42 | |||
| Hard, 8k tokens per step (budget sweep) | Reactive (thinking off) | 0.42 | |||
| Hard, 16k tokens per step (budget sweep) | Reactive (thinking off) | 0.42 | |||
| Hard, 32k tokens per step (budget sweep) | Reactive (thinking off) | 0.42 | |||
| Hard, 4k tokens per step (budget sweep) | AgileThinker | 0.39 | |||
| Hard, 8k tokens per step (budget sweep) | Planning (thinking on) | 0.06 | |||
| Hard, 4k tokens per step (budget sweep) | Planning (thinking on) | 0.00 |
Scores at each effort level
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: The Real-Time Reasoning Gym paper
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| Hard, 32k tokens per step (budget sweep) | Planning (thinking on) | 0.81 | |||
| Hard, 32k tokens per step (budget sweep) | AgileThinker | 0.77 | |||
| Hard, 16k tokens per step (budget sweep) | AgileThinker | 0.74 | |||
| Hard, 8k tokens per step (budget sweep) | AgileThinker | 0.55 | |||
| Hard, 16k tokens per step (budget sweep) | Planning (thinking on) | 0.44 | |||
| Hard, 4k tokens per step (budget sweep) | Reactive (thinking off) | 0.42 | |||
| Hard, 8k tokens per step (budget sweep) | Reactive (thinking off) | 0.42 | |||
| Hard, 16k tokens per step (budget sweep) | Reactive (thinking off) | 0.42 | |||
| Hard, 32k tokens per step (budget sweep) | Reactive (thinking off) | 0.42 | |||
| Hard, 4k tokens per step (budget sweep) | AgileThinker | 0.39 | |||
| Hard, 8k tokens per step (budget sweep) | Planning (thinking on) | 0.06 | |||
| Hard, 4k tokens per step (budget sweep) | Planning (thinking on) | 0.00 |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The Real-Time Reasoning Gym paper
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official Real-Time Reasoning Gym 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 11 months ago, in December 2025, for DeepSeek-V3.2. 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 benchmark’s paper (3 entries).
- Note Models scored Only 1 model we track, plus 2 other entries on its leaderboard.
- Warning Newest score The newest came 11 months ago, in December 2025, for DeepSeek-V3.2.
- Note Frontier models None of the top 10 has a score. The strongest tested is DeepSeek-V3.2 (#91).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on arxiv.org: 12 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.