Pokémon SVG Bench
Pokémon SVG Bench evaluates a model's knowledge of Pokémon and its ability to generate SVGs.
Learn moreTest cases
15
Models tested
39
X: SVG anchors Y: Visual Score
|V1.1 |
| Rank | Model | ||||
|---|---|---|---|---|---|
 | Gemini 3.6 Flash Official API. thinking_level: high | 50.72 | 62.40 | 48.00 | 48.60 |
 | Gemini 3.1 Pro Official API. reasoning_effort: medium | 41.44 | 59.60 | 49.20 | 31.40 |
 | Gemini 3.5 Flash Official API. reasoning_effort: medium | 40.57 | 57.40 | 40.40 | 35.40 |
| 4 | Arrow 1.1 Official API | 40.45 | 37.00 | 55.60 | 33.00 |
| 5 | GPT-5.6 Sol Cloudflare Proxy API. Reasoning_effortt: medium | 39.94 | 63.60 | 43.40 | 30.60 |
| 6 | Claude Fable 5 Cloudflare Proxy API. Thinking: adaptive. Effort: high | 37.88 | 53.20 | 48.00 | 27.40 |
| 7 | GPT-5.5 Cloudflare Proxy API. Reasoning_effortt: medium | 34.54 | 54.00 | 38.20 | 26.40 |
| 8 | Kimi K3 Official API. Reasoning_effortt: max | 32.83 | 53.40 | 37.80 | 23.60 |
| 9 | Claude Opus 5 Cloudflare Proxy API. Thinking: adaptive. Effort: high | 32.49 | 54.20 | 38.00 | 22.60 |
| 10 | Gemini 3.5 Flash-Lite Official API. thinking_level: high | 32.06 | 43.80 | 32.40 | 28.20 |
| 11 | Gemini 3 Flash Official API. reasoning_effort: minimal | 27.33 | 46.80 | 35.60 | 16.60 |
| 12 | Qwen3.7-Max Official API. Thinking mode enabled. | 26.99 | 34.60 | 30.20 | 22.80 |
| 13 | Claude Opus 4.8 Cloudflare Proxy API. Thinking: adaptive. Effort: high | 26.86 | 42.80 | 30.20 | 20.00 |
| 14 | GPT-5.6 Terra Cloudflare Proxy API. Reasoning_effortt: medium | 25.43 | 35.60 | 29.80 | 19.80 |
| 15 | Doubao-Seed-2.1-pro Official API | 23.75 | 32.80 | 25.40 | 20.00 |
| 16 | Grok 4.5 Official API | 22.64 | 32.60 | 25.00 | 18.20 |
| 17 | GLM-5.2 Official API | 22.55 | 37.80 | 23.60 | 17.20 |
| 18 | Qwen3.6-Max-Preview Official API. Thinking mode enabled. | 22.05 | 24.40 | 25.80 | 19.20 |
| 19 | Claude Opus 4.7 Cloudflare Proxy API. Thinking: adaptive. Effort: medium | 21.67 | 31.20 | 25.40 | 16.60 |
| 20 | Claude Sonnet 4.6 Cloudflare Proxy API. Effort: medium | 21.23 | 30.20 | 22.00 | 18.00 |
| 21 | GLM-5.1 Official API | 20.79 | 33.00 | 23.60 | 15.40 |
| 22 | Qwen3.7-Plus Official API. Thinking mode enabled. | 19.32 | 22.40 | 22.80 | 16.40 |
| 23 | Hy3 OpenRouter API | 18.75 | 27.00 | 22.60 | 14.00 |
| 24 | MiMo-V2.5-Pro Official API | 18.66 | 27.60 | 23.40 | 13.20 |
| 25 | MiMo-V2.5 Official API | 17.90 | 22.80 | 21.40 | 14.40 |
| 26 | DeepSeek v4 Pro Official API. Thinking mode enabled. reasoning_effort: high | 17.76 | 29.20 | 18.80 | 13.60 |
| 27 | Qwen3.6-Plus Official API. Thinking mode enabled. | 17.54 | 15.40 | 23.60 | 14.80 |
| 28 | Doubao-Seed-2.0-pro Official API | 17.39 | 27.40 | 24.60 | 10.20 |
| 29 | Inkling OpenRouter API | 15.63 | 20.60 | 18.60 | 12.40 |
| 30 | Composer 2.5 Generated by Cursor Subagents | 15.61 | 26.00 | 23.00 | 8.20 |
| 31 | Composer 2 Generated by Cursor Subagents | 14.13 | 21.00 | 21.20 | 8.00 |
| 32 | Grok 4.3 Official API | 13.82 | 20.00 | 15.40 | 11.00 |
| 33 | DeepSeek v4 Flash Official API. Thinking mode disabled. | 13.25 | 19.80 | 16.80 | 9.20 |
| 34 | Step 3.7 Flash OpenRouter API | 10.03 | 19.40 | 12.00 | 6.00 |
| 35 | Hy3 preview OpenRouter API | 10.01 | 19.20 | 15.60 | 4.00 |
| 36 | Gemini 3.1 Flash-Lite Official API. reasoning_effort: minimal | 9.05 | 28.00 | 11.40 | 1.80 |
| 37 | Kimi K2.6 Official API. Thinking mode enabled. | 7.55 | 21.00 | 6.40 | 4.00 |
| 38 | Step 3.5 Flash OpenRouter API | 3.93 | 10.20 | 4.60 | 1.60 |
| 39 | Laguna S 2.1 OpenRouter API | 0.47 | 2.80 | 0.00 | 0.00 |