Mercor publishes APEX results for Inkling 2 Small, up 13 points on Inkling on APEX-Agents

Mercor, which says it helped Thinking Machines evaluate Inkling 2 Small ahead of its launch, reports the model's scores on its APEX leaderboards:

• APEX-Agents: 46.8% Pass@1 and a 60.9% mean score, 13 points above Inkling (33.8% Pass@1, launched in July). By domain, Pass@1 is 62.2% in Law (+12.5), 40.3% in Management Consulting (+14.8) and 37.8% in Investment Banking (+11.5).
• APEX-Accounting: 14.5% Pass@1 and a 62.0% mean score. Mercor calls it the second-highest-scoring open model there, behind Kimi K3.1 Pro (62.4% mean score).

These are Mercor's own figures.