Benchmark Almanac

Slay the Spire Bench (alavico)

Claude Sonnet 4.6 51.0 leads since June 2026
Not in model cards

Models choose an action from the text state of Slay the Spire. The campaign report measures floors reached and wins at each difficulty.

The maker reports 12 language models and two automated baselines across 37 runs. All use Ironclad, with two seeds and three difficulty settings across runs.

Made by
alavico
Released
June 2026
Kind
Video games
Level
Everyday
Scored
Checked automatically
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Actions
Interaction
Agentic
Environment
Slay the Spire
Items from
Generated
Refresh
One fixed set
Run by
Anyone
In views
Games, Agents
Strongest model tested
Claude Opus 5 #11 on the capability index
Last model added
July 2026
Scores refreshed
November 20th, 2026
14
Entries on its own leaderboard, 12 models we track

5 models share the top score of 51.0 among models released by July 2026 5 models share the top score of 51.0 among models released by July 2026

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.

The newest model with a score here came out in July 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 Claude Sonnet 4.6 Ascension 0, medium effort 51.0
  2. 2 GPT-5.4 Mini Ascension 0, medium effort 51.0
  3. 3 Kimi K2.6 Ascension 0, medium effort 51.0
  4. 4 GPT-5.5 Ascension 0, medium effort 51.0
  5. 5 GPT-5.6 Sol Ascension 0, medium effort 51.0
  6. 6 GPT-5.6 Terra Ascension 5, medium effort 50.0
  7. 7 Claude Opus 5 Ascension 0, medium effort 50.0
  8. 8 Gemini 3.5 Flash Ascension 0, medium effort 42.0
  9. 9 GLM-5.2 Ascension 0, medium effort 42.0
  10. 10 Claude Sonnet 5 Ascension 0, medium effort 40.5
  11. 11 GPT-5.6 Luna Ascension 0, medium effort 32.3
  12. 12 Gemini 3.1 Pro Ascension 0, medium effort 12.0
  13. 13 scripted Ascension 0 11.0
  14. 14 random Ascension 0 7.00

Source: Official Slay the Spire Bench (alavico) leaderboard

#EntryScoreSettingDateSource
1 Claude Sonnet 4.6 51.0 Ascension 0, medium effort, floor Feb 2026 Official leaderboard
2 GPT-5.4 Mini 51.0 Ascension 0, medium effort, floor Mar 2026 Official leaderboard
3 Kimi K2.6 51.0 Ascension 0, medium effort, floor Apr 2026 Official leaderboard
4 GPT-5.5 51.0 Ascension 0, medium effort, floor Apr 2026 Official leaderboard
5 GPT-5.6 Sol 51.0 Ascension 0, medium effort, floor Jul 2026 Official leaderboard
6 GPT-5.6 Terra 50.0 Ascension 5, medium effort, floor Jul 2026 Official leaderboard
7 Claude Opus 5 50.0 Ascension 0, medium effort, floor Jul 2026 Official leaderboard
8 Gemini 3.5 Flash 42.0 Ascension 0, medium effort, floor May 2026 Official leaderboard
9 GLM-5.2 42.0 Ascension 0, medium effort, floor Jun 2026 Official leaderboard
10 Claude Sonnet 5 40.5 Ascension 0, medium effort, floor Jun 2026 Official leaderboard
11 GPT-5.6 Luna 32.3 Ascension 0, medium effort, floor Jul 2026 Official leaderboard
12 Gemini 3.1 Pro 12.0 Ascension 0, medium effort, floor Feb 2026 Official leaderboard
13 scripted 11.0 Ascension 0, scripted Official leaderboard
14 random 7.00 Ascension 0, random Official leaderboard
#ModelScoreSettingDateSource
1 Claude Sonnet 4.6 51.0 Ascension 0, medium effort, floor Feb 2026 Official leaderboard
2 GPT-5.4 Mini 51.0 Ascension 0, medium effort, floor Mar 2026 Official leaderboard
3 Kimi K2.6 51.0 Ascension 0, medium effort, floor Apr 2026 Official leaderboard
4 GPT-5.5 51.0 Ascension 0, medium effort, floor Apr 2026 Official leaderboard
5 GPT-5.6 Sol 51.0 Ascension 0, medium effort, floor Jul 2026 Official leaderboard
6 GPT-5.6 Terra 50.0 Ascension 5, medium effort, floor Jul 2026 Official leaderboard
7 Claude Opus 5 50.0 Ascension 0, medium effort, floor Jul 2026 Official leaderboard
8 Gemini 3.5 Flash 42.0 Ascension 0, medium effort, floor May 2026 Official leaderboard
9 GLM-5.2 42.0 Ascension 0, medium effort, floor Jun 2026 Official leaderboard
10 Claude Sonnet 5 40.5 Ascension 0, medium effort, floor Jun 2026 Official leaderboard
11 GPT-5.6 Luna 32.3 Ascension 0, medium effort, floor Jul 2026 Official leaderboard
12 Gemini 3.1 Pro 12.0 Ascension 0, medium effort, floor Feb 2026 Official leaderboard

The best score, 51.0, costs $3.50 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 Cost frontier One model at each effort level
0.00 20.0 40.0 60.0 $0.10 $1.00 $10 $100 Cost per task, log scale GPT-5.4 Mini, Ascension 0, medium effort GPT-5.6 Terra, Ascension 0, medium effort GPT-5.6 Luna, Ascension 0, medium effort 0.00 20.0 40.0 60.0 $0.10 $1.00 $10 $100 Cost per task, log scale GPT-5.4 Mini, Ascension 0, medium effort GPT-5.6 Terra, Ascension 0, medium effort

Source: Official Slay the Spire Bench (alavico) leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Claude Sonnet 4.6 Ascension 0, medium effort floor 51.0 $22 632,639
GPT-5.4 Mini Ascension 0, medium effort floor 51.0 $3.50 557,850
Kimi K2.6 Ascension 0, medium effort floor 51.0 $9.67 1,307,901
GPT-5.5 Ascension 0, medium effort floor 51.0 $11 165,440
GPT-5.5 Ascension 5, medium effort floor 51.0 $10 152,798
GPT-5.5 Ascension 10, medium effort floor 51.0 $13 184,518
GPT-5.6 Sol Ascension 0, medium effort floor 51.0 $5.34 49,501
GPT-5.6 Sol Ascension 5, medium effort floor 51.0 $5.96 58,323
GPT-5.6 Sol Ascension 10, medium effort floor 51.0 $6.63 66,349
GPT-5.6 Terra Ascension 5, medium effort floor 50.0 $2.63 65,213
Claude Opus 5 Ascension 0, medium effort floor 50.0 $5.94 85,107
Claude Opus 5 Ascension 10, medium effort floor 50.0 $7.96 122,536
GPT-5.6 Terra Ascension 0, high effort floor 45.0 $2.97 83,679
Gemini 3.5 Flash Ascension 0, medium effort floor 42.0 $18 174,989
GLM-5.2 Ascension 0, medium effort floor 42.0 $3.12 448,414
GPT-5.6 Terra Ascension 0, medium effort floor 41.5 $1.80 42,699
GPT-5.4 Mini Ascension 5, medium effort floor 41.0 $2.71 442,589
Claude Sonnet 5 Ascension 0, medium effort floor 40.5 $2.85 94,902
Kimi K2.6 Ascension 10, medium effort floor 33.0 $4.25 752,778
Claude Opus 5 Ascension 5, medium effort floor 33.0 $4.84 73,080
GPT-5.6 Luna Ascension 0, medium effort floor 32.3 $0.11 30,515
GPT-5.4 Mini Ascension 0, low effort floor 17.0 $1.26 31,827
GPT-5.4 Mini Ascension 10, medium effort floor 16.0 $0.82 119,485
GPT-5.6 Terra Ascension 10, medium effort floor 14.0 $0.45 11,236
Gemini 3.1 Pro Ascension 0, medium effort floor 12.0 $2.86 32,848

The best score, 51.0, costs $3.50 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 Cost frontier One model at each effort level
0.00 20.0 40.0 60.0 $0.10 $1.00 $10 $100 Cost per task, log scale GPT-5.4 Mini, Ascension 0, medium effort GPT-5.6 Terra, Ascension 0, medium effort GPT-5.6 Luna, Ascension 0, medium effort 0.00 20.0 40.0 60.0 $0.10 $1.00 $10 $100 Cost per task, log scale GPT-5.4 Mini, Ascension 0, medium effort GPT-5.6 Terra, Ascension 0, medium effort

Source: Official Slay the Spire Bench (alavico) leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Claude Sonnet 4.6 Ascension 0, medium effort floor 51.0 $22 632,639
GPT-5.4 Mini Ascension 0, medium effort floor 51.0 $3.50 557,850
Kimi K2.6 Ascension 0, medium effort floor 51.0 $9.67 1,307,901
GPT-5.5 Ascension 0, medium effort floor 51.0 $11 165,440
GPT-5.5 Ascension 5, medium effort floor 51.0 $10 152,798
GPT-5.5 Ascension 10, medium effort floor 51.0 $13 184,518
GPT-5.6 Sol Ascension 0, medium effort floor 51.0 $5.34 49,501
GPT-5.6 Sol Ascension 5, medium effort floor 51.0 $5.96 58,323
GPT-5.6 Sol Ascension 10, medium effort floor 51.0 $6.63 66,349
GPT-5.6 Terra Ascension 5, medium effort floor 50.0 $2.63 65,213
Claude Opus 5 Ascension 0, medium effort floor 50.0 $5.94 85,107
Claude Opus 5 Ascension 10, medium effort floor 50.0 $7.96 122,536
GPT-5.6 Terra Ascension 0, high effort floor 45.0 $2.97 83,679
Gemini 3.5 Flash Ascension 0, medium effort floor 42.0 $18 174,989
GLM-5.2 Ascension 0, medium effort floor 42.0 $3.12 448,414
GPT-5.6 Terra Ascension 0, medium effort floor 41.5 $1.80 42,699
GPT-5.4 Mini Ascension 5, medium effort floor 41.0 $2.71 442,589
Claude Sonnet 5 Ascension 0, medium effort floor 40.5 $2.85 94,902
Kimi K2.6 Ascension 10, medium effort floor 33.0 $4.25 752,778
Claude Opus 5 Ascension 5, medium effort floor 33.0 $4.84 73,080
GPT-5.6 Luna Ascension 0, medium effort floor 32.3 $0.11 30,515
GPT-5.4 Mini Ascension 0, low effort floor 17.0 $1.26 31,827
GPT-5.4 Mini Ascension 10, medium effort floor 16.0 $0.82 119,485
GPT-5.6 Terra Ascension 10, medium effort floor 14.0 $0.45 11,236
Gemini 3.1 Pro Ascension 0, medium effort floor 12.0 $2.86 32,848

Number one over time

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

Anthropic
Own leaderboard Not tested since Jul 2026

Source: Official Slay the Spire Bench (alavico) 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
10.0 20.0 30.0 40.0 50.0 60.0 2026 10.0 20.0 30.0 40.0 50.0 60.0 2026

Source: Official Slay the Spire Bench (alavico) 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 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (14 entries).
  • Pass Models scored 12 models we track, plus 2 other entries on its leaderboard.
  • Pass Newest score The newest is from July 2026, for Claude Opus 5.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Claude Opus 5 (#11).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on alavico.github.io: 25 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.

How we check