Key facts
- Microsoft CEO Satya Nadella said Microsoft-Decision-1.1 is rolling out for wide use on Azure immediately.
- Sqwish, a UK-based AI startup, announced a $3 million pre-seed round for its kettei fat decision model.
- TypeSafe became a $7.5 billion company after releasing its Jev family of models, which dominates usage.
- OpenRouter handled more than 1.5 billion queries to TypeSafe's models in the week before the report.
- Sqwish cofounders Ushnish Sengupta and Federica Freddi met as Cambridge graduate students.
- Sqwish offers built-in post-training tools that use a customer's own data to improve the model's judgement.
Microsoft CEO Satya Nadella said on November 17, 2026, that the company is rolling out its decision model, Microsoft-Decision-1.1, for wide use on Azure immediately. Decision models are large language models that do not produce output tokens; instead, they are tuned to decide between preset options quickly and cheaply while registering confidence in the choice. They are used in production environments where fast decisions can matter more than the reasoning of a slower, more expensive frontier model.
In October, more than one in four requests on Vercel's AI model router went to decision models, according to TechCrunch. TypeSafe became a $7.5 billion company after releasing its Jev family of models, which continues to dominate usage on services like OpenRouter, which handled more than 1.5 billion queries to TypeSafe's models in the week before the report, TechCrunch said.
Sqwish, a UK-based AI startup behind a decision model called kettei fat, announced a $3 million pre-seed round on November 17. Cofounders Ushnish Sengupta and Federica Freddi, who met as Cambridge graduate students, had been developing small LLMs when Jev arrived, and they realized their architectures could be repurposed as a decision model. Sqwish's offering includes built-in post-training tools that use a customer's own data to improve the model's judgement.
TypeSafe CEO Diogo Almedia told TechCrunch that the hardest part of building Jev was finely calibrating its judgement using statistically-adjusted synthetic data. Sengupta told TechCrunch that what makes Sqwish not a commodity is that customers can keep uploading their production data so their models keep improving on the tasks they care about.
Ranking decision models is difficult. Sqwish's kettei fat has performed well on some benchmarks, but most builders say the real proof is how a model performs in a specific context. Achint Srivastava, the Microsoft engineering VP who led the Decision-1.1 project, wrote in an email that models overly focused on benchmarks fall flat under real enterprise usage quickly. Freddi predicted that future systems will see decision models working alongside frontier LLMs, and said that even in that future, decision models could be a $10 billion to $50 billion opportunity.
