IOI 2026 (Vals v2)
Code
The six IOI 2026 problems under Vals' v2 OpenCode protocol without interactive grading feedback. Final solutions receive official subtask points, averaged across the problems.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 4 Argon | 100.0 |
| GPT-5.6 Terra | 100.0 |
| GPT-6 Astra | 100.0 |
| GPT-6.1 Sol | 90.7 |
| GPT-5.6 Sol | 85.2 |
| Muse Wonder Full | 78.7 |
| Claude Sonnet 5.5 | 75.8 |
| Kimi K3.1 Pro | 70.2 |
| GPT-6 Sol | 68.5 |
| Claude Opus 5 | 65.8 |
| Gemini 3.7 Flash | 60.8 |
| Qwen3.8-Max | 60.2 |
| Muse Wonder | 58.5 |
| GPT-6 Luna | 58.3 |
| Kimi K3 | 57.8 |
| DeepSeek V4 Pro 0813 | 56.3 |
| Grok 4.7 | 55.3 |
| Hy4 Preview | 55.3 |
| Gemini 3.8 Flash | 53.8 |
| GLM-5.3 | 53.5 |
| Gemini 3.1 Pro Preview | 53.2 |
| Muse Spark 1.3 | 52.5 |
| GPT-5.6 Luna | 50.5 |
| Mistral Large 4 Preview | 47.8 |
| GLM-5.3 Flash | 47.7 |
| Grok 4.6 | 43.5 |
| DeepSeek V4.1 Flash | 43.0 |
| Claude Haiku 5.5 | 42.8 |
| Sonnet 5 | 42.5 |
| GPT-5.3-Codex | 41.7 |
| Grok 4.5 | 41.0 |
| Ember-1 | 39.3 |
| MiMo-V2.6-Flash | 38.3 |
| Beam 1.0 501B A23B | 37.7 |
| Step 5 Preview | 37.7 |
| Qwen3.8 27B | 36.8 |
| DeepSeek-V4-Flash-0731 | 36.0 |
| MiMo-V2.6-Pro | 35.2 |
| Gemini 3.6 Flash | 23.7 |
| Muse Spark 1.2 | 20.2 |