Benchmark Almanac

mage-bench

GPT-5.4 Mini 1617.0 leads since March 2026
Not in model cards

Models play Magic: The Gathering in a benchmark that compares strategic play across games.

The public site reports a comparison of 34 models in Magic: The Gathering. The best score is not stated on the checked page.

Made by
mage-bench
Released
January 2026
Kind
Multi-agent games
Level
Expert
Scored
Settled by events
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Actions
Interaction
Multi-agent
Environment
Magic: The Gathering
Items from
Live play or votes
Refresh
Live
Run by
Anyone
In views
Games, Agents, Kept fresh
Strongest model tested
GPT-5.4 Mini #74 on the capability index
Last model added
March 2026
Scores refreshed
November 20th, 2026
11
Models scored on its own leaderboard

GPT-5.4 Mini leads with 1617.0 among models released by March 2026

Best score per model 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.

The newest model with a score here came out in March 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 GPT-5.4 Mini medium 1617.0
  2. 2 Mistral Medium 3.1 1616.0
  3. 3 DeepSeek-V3.2 1616.0
  4. 4 GPT-5.3 Codex medium 1616.0
  5. 5 MiniMax-M2.5 medium 1616.0
  6. 6 Gemini 3.1 Flash-Lite 1599.0
  7. 7 Qwen3-235B-A22B-Instruct Jul 2025 1584.0
  8. 8 Grok 4 Fast medium 1584.0
  9. 9 Qwen3-Max low 1584.0
  10. 10 MiMo-V2-Flash medium 1584.0
  11. 11 GPT-5.4 Nano low 1584.0

Source: Official mage-bench leaderboard

#ModelScoreSettingDateLinkSource
1 GPT-5.4 Mini 1617.0 medium Mar 2026 mage-bench.com Official leaderboard
2 Mistral Medium 3.1 1616.0 Aug 2025 mage-bench.com Official leaderboard
3 DeepSeek-V3.2 1616.0 Dec 2025 mage-bench.com Official leaderboard
4 GPT-5.3 Codex 1616.0 medium Feb 2026 mage-bench.com Official leaderboard
5 MiniMax-M2.5 1616.0 medium Feb 2026 mage-bench.com Official leaderboard
6 Gemini 3.1 Flash-Lite 1599.0 Mar 2026 mage-bench.com Official leaderboard
7 Qwen3-235B-A22B-Instruct Jul 2025 1584.0 Jul 2025 mage-bench.com Official leaderboard
8 Grok 4 Fast 1584.0 medium Sep 2025 mage-bench.com Official leaderboard
9 Qwen3-Max 1584.0 low Sep 2025 mage-bench.com Official leaderboard
10 MiMo-V2-Flash 1584.0 medium Dec 2025 mage-bench.com Official leaderboard
11 GPT-5.4 Nano 1584.0 low Mar 2026 mage-bench.com Official leaderboard

The best score, 1617.0, costs $1.28 a task

Runs on the cost frontier score higher than any run that's cheaper. GPT-5.4 Nano at low comes within 33 of the best for $0.08 a task, 16 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

MiniMax OpenAI Cost frontier One model at each effort level
0.00 500.0 1000.0 1500.0 2000.0 $0.01 $0.10 $1.00 $10 Cost per task, log scale GPT-5.4 Mini, medium MiniMax-M2.5, medium 0.00 500.0 1000.0 1500.0 2000.0 $0.01 $0.10 $1.00 $10 Cost per task, log scale GPT-5.4 Mini, medium

Source: Official mage-bench leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5.4 Mini medium 1617.0 $1.28
Mistral Medium 3.1 1616.0 $0.24
DeepSeek-V3.2 1616.0 $0.55
GPT-5.3 Codex medium 1616.0 $1.77
MiniMax-M2.5 medium 1616.0 $0.19
Gemini 3.1 Flash-Lite 1599.0 $0.30
Qwen3-235B-A22B-Instruct Jul 2025 1584.0 $0.11
Grok 4 Fast medium 1584.0 $0.30
Qwen3-Max low 1584.0 $0.57
MiMo-V2-Flash medium 1584.0 $0.20
GPT-5.4 Nano low 1584.0 $0.08

2 models have held number one on the official leaderboard since its release in January 2026

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

Mistral AI OpenAI
Own leaderboard Not tested since Mar 2026
GPT-5.4 Mini 1617.0

Source: Official mage-bench leaderboard

Scores over time

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
1580.0 1590.0 1600.0 1610.0 1620.0 2026 1580.0 1590.0 1600.0 1610.0 1620.0 2026

Source: Official mage-bench 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

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 8 months ago, in March 2026, for GPT-5.4 Mini. 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 (11 entries).
  • Pass Models scored 11 models we track.
  • Warning Newest score The newest came 8 months ago, in March 2026, for GPT-5.4 Mini.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.4 Mini (#74).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on mage-bench.com: 11 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