Atlas

Benchmarks

← All benchmarks

IOI 2024 (Vals v2)

Code

The six IOI 2024 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)

ModelScore
Claude Fable 5.1100.0
Claude Opus 5.5100.0
Claude Opus 5100.0
GPT-5.6 Sol100.0
GPT-6 Astra100.0
GPT-6 Sol100.0
GPT-5.6 Luna85.0
GPT-5.6 Terra85.0
Qwen3.8-Max78.2
GLM-5.376.3
Gemini 3.7 Flash72.8
GPT-6 Luna69.3
GPT-5.3-Codex62.8
Grok 4.761.7
Kimi K359.7
Muse Spark 1.356.0
Gemini 3.8 Flash55.7
Hy4 Preview54.2
GLM-5.3 Flash53.2
Sonnet 550.5
Grok 4.650.3
DeepSeek V4 Pro 081350.0
Grok 4.547.2
MiMo-V2.6-Pro43.5
DeepSeek V4.1 Flash42.7
Gemini 3.6 Flash40.7
DeepSeek-V4-Flash-073134.0
Qwen3.8 27B31.8
Muse Spark 1.222.0
Inkling-Small17.2
Inkling4.2
Mercury 2.52.7
Loading Atlas data…