Inkling 2 Mini: benchmark results

Thinking Machines Lab's Inkling 2 Mini (Nov 20, 2026), an Apache 2.0 multimodal looped-transformer MoE (26.4B total, 3.4B unique active / 5.4B effective active per token across 3 looped passes; mixed NVFP4/MXFP8). Provider: Thinking Machines. Released 2026-11-20. Access: Open.

Aggregate Rating 1745 ± 14, rank #187 of 10516 rated models, from 123 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AA-AnalystAgent - Pass@K91.25Pass@K (%)100
LLM Stats (CritPT)31.4Score (%)100
AA-AnalystAgent - Pass@171Pass@1 (%)98.4
LLM Stats (LongBench v2)66.9Score (%)94.7
LLM Stats (ScreenSpot Pro)76.5Score (%)87.5
LLM Stats (MCP Atlas)82.7Score (%)82.2
Vals AI CyberBench74.58Accuracy (%)81.5
AA-Briefcase - Presentation Elo1440.13Elo76.8
Vals AI MedScribe86.37Accuracy (%)76.8
AA GDP.pdf - Legal - Criterion Pass Rate81.74Criterion Pass Rate (%)74.6
EnterpriseOps-Gym-AA - Email63.46Task Success (%)74.2
EnterpriseOps-Gym-AA - ITSM36.46Task Success (%)71

Interactive version: theaggregate.ai/model?slug=inkling-2-mini · How It Works · Data refreshed daily, snapshot 2026-11-21.