Task results
Winning a card game at its hardest setting (SpireClimb)
The model plays both Slay the Spire games at their hardest settings by operating the screen.
How to read this table
- Higher rows are rated higher. A range like “1–2” means the gap is within the margin of error.
- The score is the share of tries on a shared set of tasks that the model completed (%).
- A gap of 3 means 3 successes in 100 tries; ± is how far the score moves when re-measured.
- Prices are a guide per million tokens, reading then writing.
- Length is how much text the model can read at once (tokens).
± range of the top model
| Rank | Model | Developer | Success rate (%) | Evaluated | Input $ /1M | Output $ /1M | Context | Supports | Weights | Released | Price and specs |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1–2 | GPT-6 Astra | OpenAI | Rating 25.7 ±7.2 | $10 | $50 | 1.05M | OpenAI Released 2026-09 Price $10 / $50 Length 1.05M | ||||
| 1–2 | GPT-6.1 Sol | OpenAI | Rating 24.7 ±7.3 | $2 | $10 | 1.05M | OpenAI Released 2026-09 Price $2 / $10 Length 1.05M | ||||
| 3–4 | GPT-6 Sol | OpenAI | Rating 7.5 ±4.2 | $2 | $10 | 1.05M | OpenAI Released 2026-09 Price $2 / $10 Length 1.05M | ||||
| 3–6 | Claude Opus 5.5 | Anthropic | Rating 3.3 ±2.8 | $4 | $20 | 1M | Anthropic Released 2026-09 Price $4 / $20 Length 1M | ||||
| 4–7 | Claude Fable 5.1 | Anthropic | Rating 2.0 ±2.2 | $10 | $50 | 1M | Anthropic Released 2026-09 Price $10 / $50 Length 1M | ||||
| 4–7 | Claude Sonnet 5.5 | Anthropic | Rating 1.3 ±1.8 | $2 | $10 | 1M | Anthropic Released 2026-09 Price $2 / $10 Length 1M | ||||
| 5–7 | GPT-6 Luna | OpenAI | Rating 0.0 ±0.0 | $0.1 | $0.5 | 1.05M | OpenAI Released 2026-09 Price $0.1 / $0.5 Length 1.05M |
Best success rate over time
2026-09-01Release date2026-09-29
How to read this page
- The value is the share of runs won to the end.
- The success rate is the share of tries that succeeded. A gap of 1 point is 1 success out of 100 tries.
- In the Supports column the icons are, from left: can use tools / can read images / can answer in a fixed format (JSON); faded means not supported or unknown.
- Open-weights models have their contents (the weights) published, so you can also run them on your own computer or server. Models without the open-weights mark are normally used through the developer's own services (apps or API); for some models the weights status is unknown.
- Names such as ChatGPT next to a model are the consumer services its developer offers. Check on the service itself whether that model is available there.
Data: Arena, Epoch AI, Swallow LLM team (CC BY 4.0), Models.dev (MIT), updated 2026-10-24, data sources