MageBench S2: leaderboard
MageBench Season 2: native combined 1v1 Elo for Magic: The Gathering agents, preserving exact source model IDs and reasoning efforts. Format-specific pools and Commander exhibitions are not max-pooled.
Metric: Combined 1v1 Elo. Source: mage-bench.com. Saturation forecast: Around May 2027. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 Mini (Medium) | 1617 | #212 (GPT-5.4 Mini) |
| 2 | DeepSeek V3.2 | 1616 | #205 |
| 3 | Mistral Medium 3.1 | 1616 | #408 |
| 4 | Gemini 3.1 Flash Lite (Preview) | 1599 | #194 |
| 5 | Qwen 3 235B A22B 2507 | 1584 | #260 |
| 6 | GPT-5.4 Nano (Low) | 1584 | #328 (GPT-5.4 Nano) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=magebench-s2 · How It Works · Data refreshed daily, snapshot 2026-10-20.