Shroombench (mushroom identification, top-5)
Multimodal · 2026-09-02
Share of Shroombench's 1,040 mushroom photos (55 Polish species from FungiTastic) for which the expert-labelled species is among the five Latin binomials the model returns, by string matching with accepted synonyms. It comes from the same answers as the top-1 headline and is a component of its factor group.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 3.6 Flash | 85.0 |
| Gemini 3.8 Flash | 85.0 |
| Gemini 3.7 Flash | 82.0 |
| Muse Spark 1.3 | 74.0 |
| GPT-6 Astra | 73.0 |
| Claude Fable 5 | 72.0 |
| GLM-5.3 Flash | 72.0 |
| Claude Fable 5.1 | 70.0 |
| Kimi K3 | 70.0 |
| Claude Opus 5 | 63.0 |
| GPT-5.6 Sol Pro | 62.0 |
| Grok 4.6 | 62.0 |
| GPT-5.6 Sol | 61.0 |
| GPT-5.6 Terra | 54.0 |
| Muse Spark 1.2 | 54.0 |
| GPT-5.6 Luna | 49.0 |
| Sonnet 5 | 47.0 |
| DeepSeek V4 Flash Vision Exp | 46.0 |
| Qwen3.8 Flash | 43.0 |
| MiniMax M3 | 40.0 |
| Qwen3.8 27B | 24.0 |