Routers that let Jev pick the model, the tool or the skill before the agent starts, with the latency and cost saving each author measured.
68 builds · page 3 of 3
Satya Nadella
@satyanadella
Introducing Microsoft-Decision-1, our new model for fast decision-making.
It delivers top performance on structured decision tasks, outperforming both LLMs and other decision models in latency and quality.
We’re already testing it across Microsoft for everything from incident response and quality control to scientific discovery.
Jev raised $870M at $7.5B yesterday. On 262 of our real decisions it beat Claude Haiku 5.5, 204 to 177.
TypeSafe's decision model got that valuation weeks after launch, in the same week Anthropic shipped Haiku 5.5 at $0.10 per million input tokens. So we checked whether a cheap general model can do the deciding. Haiku 5.5 got the labeled holdout cases from 17 decision systems we run every day, with the same instructions, word for word.
→ Right picks: Jev 204/262, Haiku 5.5 177/262
→ By system: Haiku won 1, tied 5, lost 11
→ Cost of the 262 calls: $0.058 on Haiku, $0.022 on Jev (2.7x)
→ Median answer: 1.16 s vs 375 ms
→ Biggest gap: "which builder gets this task" 6/12 vs 10/12
The twist: the one system Haiku won is the one that asks for judgment. "Did the agent really finish the task": Haiku 11/11, Jev 9/11.
One caveat: those instructions were tuned on Jev for weeks. Haiku read them cold.
Would you put a general model or a decision model behind your agent's routing?
Repost this now, because "Haiku is cheap enough for everything" is turning into a default nobody measured.
Follow @stas_sorokin_ for numbers from AI systems running in real production.
i just put Jev inside an open agent harness.
(150x cheaper decision-making than LLM)
agents burn a lot of tokens on decisions that are really just picking one option from a short list, like matching a client request to the right GitHub issue.
so i built a weekly client follow-up agent that does exactly that. i tested it on 40 requests, and it was as accurate as letting Opus make every match, at 150x lower decision cost.
here's the tech stack i used:
→ TrueForge as the agent harness
→ Jev for the matching decisions
→ Daytona as the sandbox for code execution
Jev is a decision model from TypeSafe AI. it takes the options from your app and returns a structured choice with a calibrated confidence score, without generating any text.
that confidence decides what gets a second look. low-confidence matches go back to Opus for review, and in my test, that review caught 5 wrong matches that would otherwise have gone through.
TrueForge is TrueFoundry's harness, and it keeps context small. it loads tool definitions only when needed, writes large tool outputs to files, and lets the agent process data in scripts instead of the conversation.
on DevRev's Enterprise Bench, it solved about as many tasks as Claude Managed Agents on roughly a third of the tokens. it also keeps credentials outside the sandbox and holds every write for approval.
if you want to try this out, TrueForge is fully open source (MIT-licensed) and runs on your own machine with a single command.
TrueForge GitHub: https://github.com/truefoundry/trueforge
(don't forget to star 🌟)
to help you reproduce my work and understand all the integration details, i wrote a detailed article. it's quoted below.