Inkling 2 Small
Thinking Machines Lab馃嚭馃嚫
thinkingmachines/inkling-2-smallInkling 2 Small is the first model in Thinking Machines Lab's second Inkling generation, an Apache 2.0 open-weight mixture-of-experts model (594B total, 17B active parameters, 1M-token context) that reasons over text and images. Trained with reinforcement learning on long-horizon coding across many agent harnesses, it targets agentic coding, autonomous research, and RAG at a low $0.30/$1 per million input/output tokens.
Cost rate
0.2x
Context
1M
Released
Nov 9, 2026
Input
TextImage
Output
Text
Support
ReasoningTool callingStructured outputs
Best at
The categories where this model ranks highest.
Business, Management, & Finance
#118 路 top 29%
Software & IT Services
#123 路 top 29%
Writing, Literature, & Language
#157 路 top 37%
Life, Physical, & Social Science
#161 路 top 38%
Performance
Median latency and throughput measured across recent requests.
Throughput105 tok/s
Latency2.34s