Reports

Trial reports rendered live from the run bundles in /runs — one JSON file per model run, written by the evaluation notebook.

• google/gemini-3.1-flash-image200 samples2026-09-10
• google/gemini-3.1-flash-image_strict200 samples2026-09-12
• minimax/minimax-m3200 samples2026-09-11
• moonshotai/kimi-k2.6200 samples2026-09-11
• moonshotai/kimi-k2.6_strict200 samples2026-09-11
• moonshotai/kimi-k3200 samples2026-09-10