Inkling scores 836 Elo on AA-Briefcase, beats DeepSeek V4 Flash

Thinking Machines Lab's Inkling scores 836 Elo on AA-Briefcase, above DeepSeek V4 Flash but below Nemotron 3 Ultra and GLM-5.2. It uses 52K output tokens per task on average, with 81 turns.

• 基准测试
• Inkling
• Thinking Machines Lab