Benchmark Almanac

Crest Arcade Mines

Gemini 4 Carbon 100% leads since October 2026
Not in model cards

Models reveal or flag cells on Minesweeper boards that can be solved without guessing. The score averages win rates across four board sizes.

Glimmer Index publishes scores for 18 distinct models. Of those, 9 completed the required set.

Made by
Glimmer Index
Kind
Puzzles and chess
Level
Everyday
Scored
Checked automatically
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Actions
Interaction
Agentic
Environment
Minesweeper
Items from
Generated
Refresh
One fixed set
Run by
Anyone
In views
Games, Agents
Strongest model tested
GPT-6.2 Astra #1 on the capability index
Last model added
November 2026
Scores refreshed
November 20th, 2026
18
Entries on its own leaderboard, 17 models we track

2 models share the top score of 100% 2 models share the top score of 100%

Best score per entry from the benchmark's own leaderboard, read on November 20th, 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. This view shows only models we track, each at its best.

  1. 1 Gemini 4 Carbon medium · Provisional 100%
  2. 2 Gemini 4 Neon medium · Provisional 100%
  3. 3 GPT-6.2 Astra medium 83.3%
  4. 4 Claude Haiku 5.5 medium 70.8%
  5. 5 Gemini 4 Hydrogen medium 70.8%
  6. 6 Claude Sonnet 5.5 medium 58.3%
  7. 7 GPT-6.1 Sol medium 58.3%
  8. 8 GPT-6 Astra medium 50%
  9. 9 GPT-6 Sol medium 41.7%
  10. 10 GPT-6.1 Terra medium 20.8%
  11. 11 Qwen3.8-Flash-Next low · Provisional 18.8%
  12. 12 Gemma 3 4B none · Provisional 0%
  13. 13 gpt-oss-20b low · Provisional 0%
  14. 14 Qwen3.5-9B none · Provisional 0%
  15. 15 Qwen 3.8 27B none · Provisional 0%
  16. 16 GPT-6 Luna medium 0%
  17. 17 Gemini 4 Argon medium · Provisional 0%
  18. 18 Ministral 3 8B none · Provisional 0%

Source: Official Crest Arcade Mines leaderboard

#EntryScore ±SettingReleasedRun onSource
1 Gemini 4 Carbon 100% 69.2 to 100 medium · Provisional Oct 2026 Nov 2026 Official leaderboard
2 Gemini 4 Neon 100% 78.8 to 100 medium · Provisional Nov 2026 Nov 2026 Official leaderboard
3 GPT-6.2 Astra 83.3% ±15 medium Nov 2026 Nov 2026 Official leaderboard
4 Claude Haiku 5.5 70.8% ±13.9 medium Oct 2026 Nov 2026 Official leaderboard
5 Gemini 4 Hydrogen 70.8% ±14.6 medium Nov 2026 Nov 2026 Official leaderboard
6 Claude Sonnet 5.5 58.3% ±14.7 medium Sep 2026 Nov 2026 Official leaderboard
7 GPT-6.1 Sol 58.3% ±14.7 medium Sep 2026 Oct 2026 Official leaderboard
8 GPT-6 Astra 50% ±15.3 medium Sep 2026 Oct 2026 Official leaderboard
9 GPT-6 Sol 41.7% ±13 medium Sep 2026 Oct 2026 Official leaderboard
10 GPT-6.1 Terra 20.8% ±13.9 medium Oct 2026 Oct 2026 Official leaderboard
11 Qwen3.8-Flash-Next 18.8% ±17.4 low · Provisional Aug 2026 Nov 2026 Official leaderboard
12 Gemma 3 4B 0% 0 to 26.7 none · Provisional Mar 2025 Nov 2026 Official leaderboard
13 gpt-oss-20b 0% 0 to 26.7 low · Provisional Aug 2025 Nov 2026 Official leaderboard
14 Qwen3.5-9B 0% 0 to 22.7 none · Provisional Feb 2026 Nov 2026 Official leaderboard
15 Qwen 3.8 27B 0% 0 to 21.2 none · Provisional Aug 2026 Nov 2026 Official leaderboard
16 GPT-6 Luna 0% 0 to 11.5 medium Sep 2026 Oct 2026 Official leaderboard
17 Gemini 4 Argon 0% 0 to 26.7 medium · Provisional Sep 2026 Nov 2026 Official leaderboard
18 Ministral 3 8B open weights 0% 0 to 25.4 none · Provisional Nov 2026 Official leaderboard
#ModelScore ±SettingReleasedRun onSource
1 Gemini 4 Carbon 100% 69.2 to 100 medium · Provisional Oct 2026 Nov 2026 Official leaderboard
2 Gemini 4 Neon 100% 78.8 to 100 medium · Provisional Nov 2026 Nov 2026 Official leaderboard
3 GPT-6.2 Astra 83.3% ±15 medium Nov 2026 Nov 2026 Official leaderboard
4 Claude Haiku 5.5 70.8% ±13.9 medium Oct 2026 Nov 2026 Official leaderboard
5 Gemini 4 Hydrogen 70.8% ±14.6 medium Nov 2026 Nov 2026 Official leaderboard
6 Claude Sonnet 5.5 58.3% ±14.7 medium Sep 2026 Nov 2026 Official leaderboard
7 GPT-6.1 Sol 58.3% ±14.7 medium Sep 2026 Oct 2026 Official leaderboard
8 GPT-6 Astra 50% ±15.3 medium Sep 2026 Oct 2026 Official leaderboard
9 GPT-6 Sol 41.7% ±13 medium Sep 2026 Oct 2026 Official leaderboard
10 GPT-6.1 Terra 20.8% ±13.9 medium Oct 2026 Oct 2026 Official leaderboard
11 Qwen3.8-Flash-Next 18.8% ±17.4 low · Provisional Aug 2026 Nov 2026 Official leaderboard
12 Gemma 3 4B 0% 0 to 26.7 none · Provisional Mar 2025 Nov 2026 Official leaderboard
13 gpt-oss-20b 0% 0 to 26.7 low · Provisional Aug 2025 Nov 2026 Official leaderboard
14 Qwen3.5-9B 0% 0 to 22.7 none · Provisional Feb 2026 Nov 2026 Official leaderboard
15 Qwen 3.8 27B 0% 0 to 21.2 none · Provisional Aug 2026 Nov 2026 Official leaderboard
16 GPT-6 Luna 0% 0 to 11.5 medium Sep 2026 Oct 2026 Official leaderboard
17 Gemini 4 Argon 0% 0 to 26.7 medium · Provisional Sep 2026 Nov 2026 Official leaderboard

Rankings by part

We read 4 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Beginner, Intermediate and Expert. GPT-6 Sol leads 1 of them, and 3 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Beginner

17 models we track have a score here.

  1. 1 GPT-6 Sol ↑8 100%
  2. 2 GPT-6.1 Sol ↑5 100%
  3. 3 Claude Haiku 5.5 ↑1 100%
  4. 4 Gemini 4 Carbon ↓3 100%
  5. 5 Gemini 4 Hydrogen 100%
  6. 6 Gemini 4 Neon ↓4 100%
  7. 7 GPT-6 Astra ↑1 83.3%
  8. 8 Claude Sonnet 5.5 ↓2 83.3%
  9. 9 GPT-6.2 Astra ↓6 83.3%
  10. 10 GPT-6.1 Terra 66.7%

The best score, 100%, costs $0.94 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.

OpenAI Anthropic Google DeepMind Cost frontier One model at each effort level
0% 20% 40% 60% 80% 100% $0.001 $0.01 $0.10 $1.00 $10 Cost per task, log scale Gemini 4 Carbon, medium · Provisional Claude Haiku 5.5, medium GPT-6.1 Terra, medium GPT-6 Luna, medium 0% 20% 40% 60% 80% 100% $0.001 $0.01 $0.10 $1.00 $10 Cost per task, log scale Gemini 4 Carbon, medium · Provisional Claude Haiku 5.5, medium

Source: Official Crest Arcade Mines leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Gemini 4 Carbon medium · Provisional 100% $0.94 89,417
Gemini 4 Neon medium · Provisional 100% $2.06 124,023
GPT-6.2 Astra medium 83.3% $1.17 15,189
Claude Haiku 5.5 medium 70.8% $0.12 224,653
Gemini 4 Hydrogen medium 70.8% $0.59 69,284
Claude Sonnet 5.5 medium 58.3% $1.07 80,837
GPT-6.1 Sol medium 58.3% $0.34 18,216
GPT-6 Astra medium 50% $1.52 15,590
GPT-6 Sol medium 41.7% $0.23 17,213
GPT-6.1 Terra medium 20.8% $0.03 2,350
Qwen3.8-Flash-Next low · Provisional 18.8% 90,919
Gemma 3 4B none · Provisional 0% 620
gpt-oss-20b low · Provisional 0% 25,733
Qwen3.5-9B none · Provisional 0% 3,717
Qwen 3.8 27B none · Provisional 0% 2,721
GPT-6 Luna medium 0% $0.005 4,426
Gemini 4 Argon medium · Provisional 0% $0.05 3,398

The best score, 100%, costs $0.94 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.

OpenAI Anthropic Google DeepMind Cost frontier One model at each effort level
0% 20% 40% 60% 80% 100% $0.001 $0.01 $0.10 $1.00 $10 Cost per task, log scale Gemini 4 Carbon, medium · Provisional Claude Haiku 5.5, medium GPT-6.1 Terra, medium GPT-6 Luna, medium 0% 20% 40% 60% 80% 100% $0.001 $0.01 $0.10 $1.00 $10 Cost per task, log scale Gemini 4 Carbon, medium · Provisional Claude Haiku 5.5, medium

Source: Official Crest Arcade Mines leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Gemini 4 Carbon medium · Provisional 100% $0.94 89,417
Gemini 4 Neon medium · Provisional 100% $2.06 124,023
GPT-6.2 Astra medium 83.3% $1.17 15,189
Claude Haiku 5.5 medium 70.8% $0.12 224,653
Gemini 4 Hydrogen medium 70.8% $0.59 69,284
Claude Sonnet 5.5 medium 58.3% $1.07 80,837
GPT-6.1 Sol medium 58.3% $0.34 18,216
GPT-6 Astra medium 50% $1.52 15,590
GPT-6 Sol medium 41.7% $0.23 17,213
GPT-6.1 Terra medium 20.8% $0.03 2,350
Qwen3.8-Flash-Next low · Provisional 18.8% 90,919
Gemma 3 4B none · Provisional 0% 620
gpt-oss-20b low · Provisional 0% 25,733
Qwen3.5-9B none · Provisional 0% 3,717
Qwen 3.8 27B none · Provisional 0% 2,721
GPT-6 Luna medium 0% $0.005 4,426
Gemini 4 Argon medium · Provisional 0% $0.05 3,398

6 models have held number one on the official leaderboard since 2023

Lab claims count from the day the card came out, independent runs from the model’s release.

Google DeepMind Qwen OpenAI Anthropic
Own leaderboard Passed 85% in Oct 2026, by Gemini 4 Carbon

Source: Official Crest Arcade Mines leaderboard

Gemini 4 Carbon on the official leaderboard passed 85% in October 2026

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

Plot with other benchmarks
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2025 2026 2027 Near the ceiling (85%) Gemini 4 Carbon Claude Haiku 5.5 Claude Sonnet 5.5 GPT-6 Astra Qwen3.8-Flash-Next Gemma 3 4B 0% 20% 40% 60% 80% 100% 2025 2026 2027 Near the ceiling (85%) Gemini 4 Carbon Claude Haiku 5.5

Source: Official Crest Arcade Mines 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 10 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (18 entries).
  • Pass Models scored 17 models we track, plus 1 other entry on its leaderboard.
  • Pass Newest score The newest is from November 2026, for Gemini 4 Neon.
  • Pass Frontier models 7 of the top 10 have a score, led by GPT-6.2 Astra (#1).
  • Note Headroom Saturated. The best is 100%, past 85% since October 2026.
Leaderboard import 3 pass
  • Pass Importer We read the board on glimmerindex.com: 17 rows for models we track and 1 other entry 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