Benchmark Almanac

GlyphBench

Not in model cards

Agents choose actions from text-rendered game boards. The standard suite contains 303 tasks across six game collections and scores agents by cumulative return.

The September 2026 study reports Qwen3.5-4B improving from 0.012 to 0.319 mean return on held-out seeds across 100 tasks after training.

Released
September 2026
Kind
Video games
Level
University
Scored
Settled by events
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Actions
Interaction
Agentic
Environment
Craftax, A simulator
Items from
Generated
Refresh
One fixed set
Run by
Anyone
In views
Games, Agents, Robotics
Strongest model tested
Qwen 3.8 27B #59 on the capability index
Last model added
August 2026
Scores refreshed
November 20th, 2026
2
Models scored in its paper

Qwen 3.8 27B leads with 0.42

Best score per model from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. The headline is the plain mean of the 303 task returns plotted in Figure 3. Where a model ran with several settings or harnesses, we keep its best.

  1. 1 Qwen 3.8 27B Zero-shot 0.42
  2. 2 Qwen3.5-4B Zero-shot -0.01

Source: The GlyphBench paper

#ModelScoreSettingDateSource
1 Qwen 3.8 27B 0.42 Zero-shot Aug 2026 Official leaderboard
2 Qwen3.5-4B -0.01 Zero-shot Feb 2026 Official leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

In the benchmark’s paper
0.00 0.20 0.40 2026 0.00 0.20 0.40 2026

Source: Official GlyphBench 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

All 261 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

All 8 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
  • Pass Independent runs Scores from the benchmark’s paper (2 entries).
  • Note Models scored Only 2 models we track.
  • Pass Newest score The newest is from August 2026, for Qwen 3.8 27B.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Qwen 3.8 27B (#59).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 2 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.

How we check