🤖 AI Summary
Thinking Machines has launched Inkling 2 Small, a proprietary AI model that scores 31 on the Artificial Analysis Intelligence Index, making it the highest-scoring open weights model developed outside China. This remarkable advancement reflects a significant improvement of 18 points over its predecessor, Inkling Small, with notable enhancements in performance across various tasks, including a near quadrupling of output tokens per task. Inkling 2 Small demonstrates substantial gains in automation and coding challenges, with scores rising from 5% to 46% on AutomationBench-AA and from 5% to 25% on FrontierCode 2.0.
The model showcases varied strengths compared to competitors, outperforming Claude Sonnet 5.5 in coding tasks while still trailing in certain creative assessments. Notably, it exhibits improved accuracy and reduced hallucination rates, dropping from 61% to 27% in hallucinations, although its overall accuracy only inches up to 34%. Priced at $1.18 per task, Inkling 2 Small is positioned as a competitive option in the market, offering robust capabilities for complex tasks while being cheaper than several other high-performing models. The planned release of its weights aims to democratize access to this cutting-edge technology, making it intriguing for the AI/ML community and further energizing competition on the Intelligence Index.
Loading comments...
login to comment
loading comments...
no comments yet