# Thinking Machines Lab releases Inkling 2 Small, an AI model that companies can fine-tune with their own data

- Link: https://wowtale.net/2026/11/07/267620/?utm_source=rss&utm_medium=rss&utm_campaign=%25ec%2594%25bd%25ed%2582%25b9%25eb%25a8%25b8%25ec%258b%25a0%25ec%25a6%2588%25eb%259e%25a9-%25ec%259e%2589%25ed%2581%25b4%25eb%25a7%2581-2-%25ec%258a%25a4%25eb%25aa%25b0-%25ea%25b3%25b5%25ea%25b0%259c50%25ec%2596%25b5-%25eb%258b%25ac%25eb%259f%25ac-%25ed%2588%25ac%25ec%259e%2590%25ec%259c%25a0
- Discussion: https://startupxo.com/en/1246/
- Posted by startupxo at 2026-11-07T12:25:51.759Z
- 0 points, 1 comment, kind story, origin curated
- Written in ko, machine translated to en

## Summary

- Thinking Machines Lab, founded by former OpenAI executive Mira Murati, has released a 2-generation AI model. It is designed so that companies can further train the model on their own data and adapt it to their own work.
- For now, the weights (the learned contents of the model) are closed, and it can be used only through the company's fine-tuning service, Tinker. The company said it will release the model under the Apache 2.0 license next week, after completing an external safety review.
- According to the company, the architecture activates only 17 billion parameters out of a total of 594 billion parameters, and it supports image input. Its Artificial Analysis Intelligence Index score is 31, which the company says far exceeds its previous model.
- In the coding evaluations of the open model comparison group chosen by the company, it ranked number 2 in 4 of 6 evaluations and never ranked number 1. On the GPU code optimization evaluation PMPP-Hard, it fell short of GPT-6 Astra and Claude Opus 5.5.

## Sources

- 와우테일: https://wowtale.net/2026/11/07/267620/?utm_source=rss&utm_medium=rss&utm_campaign=%25ec%2594%25bd%25ed%2582%25b9%25eb%25a8%25b8%25ec%258b%25a0%25ec%25a6%2588%25eb%259e%25a9-%25ec%259e%2589%25ed%2581%25b4%25eb%25a7%2581-2-%25ec%258a%25a4%25eb%25aa%25b0-%25ea%25b3%25b5%25ea%25b0%259c50%25ec%2596%25b5-%25eb%258b%25ac%25eb%259f%25ac-%25ed%2588%25ac%25ec%259e%2590%25ec%259c%25a0

## Comments (1)

- **startupxo** (editorial), 0 points, 2026-11-07T12:25:51.759Z, #c519
  O: If you're a developer choosing a financial news classification model, I suggest training on your work data with Tinker and then comparing its accuracy with existing models.
  X: Across 6 coding evaluations of the open models selected by Thinking Machines Lab for comparison, Inkling 2 Small ranked 2 in 4 evaluations and never ranked 1.

JSON: https://startupxo.com/api/v1/item.json?id=1246&lang=en