# Jev

Decision models are usually judged by how well they act, yet the same models are pulled toward an easier subgoal when acting on ALFWorld. Used instead as a teacher that sees the whole trajectory after an episode and is asked whether each action helped, JEV raises a Qwen2.5-1.5B policy trained with GRPO from 72.8% to 82.9% success on ALFWorld and from 56.8% to 71.4% on WebShop, while JEV as an actor completes only 42.9% of ALFWorld tasks.

Decision models are usually judged by how well they act, yet the same models are pulled toward an easier subgoal when acting on ALFWorld. Used instead as a teacher that sees the whole trajectory after an episode and is asked whether each action helped, JEV raises a Qwen2.5-1.5B policy trained with GRPO from 72.8% to 82.9% success on ALFWorld and from 56.8% to 71.4% on WebShop, while JEV as an actor completes only 42.9% of ALFWorld tasks.

OpenRouter compared three ways to get reliable JSON out of models, using 24 messy support messages across nine models that support structured outputs: send a strict JSON Schema, force a single tool call, or switch to a decision model that only returns labels. Across the nine models, 209 of 216 replies matched a strict schema, and all 7 misses came from one provider that ignores the schema; asking for JSON only in the prompt produced 18 failures, each from a code fence. For GLM 5.3 FlashX, which only supports tools, forcing the tool call gave 22 valid outputs out of 24, against 0 when the schema was sent directly.

OpenRouter compared three ways to get reliable JSON out of models, using 24 messy support messages across nine models that support structured outputs: send a strict JSON Schema, force a single tool call, or switch to a decision model that only returns labels. Across the nine models, 209 of 216 replies matched a strict schema, and all 7 misses came from one provider that ignores the schema; asking for JSON only in the prompt produced 18 failures, each from a code fence. For GLM 5.3 FlashX, which only supports tools, forcing the tool call gave 22 valid outputs out of 24, against 0 when the schema was sent directly.

TypeSafe released Citation Telephone, a free tool that takes a cited sentence, opens the papers it cites one by one, finds the passage the sentence rests on, and labels each hop as matching, partly matching, going further or changed. On a held-out third set of 39 full-text citation pairs it found a supporting passage in 38 (97%) and reported support in 0 of 21 pairs where reviewers judged the paper did not state the claim; but hop labels agreed with reviewers in only 47% to 55% of pairs, so a warning should be treated as a reason to read the passage yourself.

TypeSafe released Citation Telephone, a free tool that takes a cited sentence, opens the papers it cites one by one, finds the passage the sentence rests on, and labels each hop as matching, partly matching, going further or changed. On a held-out third set of 39 full-text citation pairs it found a supporting passage in 38 (97%) and reported support in 0 of 21 pairs where reviewers judged the paper did not state the claim; but hop labels agreed with reviewers in only 47% to 55% of pairs, so a warning should be treated as a reason to read the passage yourself.

TypeSafe AI's probabilistic decision model Jev is now available on Vercel AI Gateway, letting an application hand in a state and get back typed Choice, Score and Boolean answers with probabilities. Regular language models generate text one token at a time for the application to parse and validate; Jev evaluates all declared questions in parallel and returns structured answers directly. TypeSafe reports Jev was up to 193.6x faster and 444.6x cheaper than LLMs on its workflow evaluations, for uses such as picking the next tool in an agent loop, deciding whether to continue or stop, scoring risk, and verifying outputs.

TypeSafe AI's probabilistic decision model Jev is now available on Vercel AI Gateway, letting an application hand in a state and get back typed Choice, Score and Boolean answers with probabilities. Regular language models generate text one token at a time for the application to parse and validate; Jev evaluates all declared questions in parallel and returns structured answers directly. TypeSafe reports Jev was up to 193.6x faster and 444.6x cheaper than LLMs on its workflow evaluations, for uses such as picking the next tool in an agent loop, deciding whether to continue or stop, scoring risk, and verifying outputs.

TypeSafe AI's decision model Jev 1.14 is now available on AI Gateway. It takes a state and a set of questions and returns typed Choice, Score and Boolean answers with probabilities instead of free-form text, answering every question in one request in parallel for classification, routing, scoring and automated verification. Version 1.14 accepts images as part of the state and adds an experimental long-context mode for documents, logs or conversation histories too long for a standard request.

TypeSafe AI's decision model Jev 1.14 is now available on AI Gateway. It takes a state and a set of questions and returns typed Choice, Score and Boolean answers with probabilities instead of free-form text, answering every question in one request in parallel for classification, routing, scoring and automated verification. Version 1.14 accepts images as part of the state and adds an experimental long-context mode for documents, logs or conversation histories too long for a standard request.

Swapping agent action selection for a fast decision model like Jev cut per-decision cost by 66.1% to 78.4% and decision time by 73.9% to 81.5%, yet task success held near baseline on ALFWorld and fell more sharply in two mobile agents, so local savings do not automatically become delivered outcomes.

Swapping agent action selection for a fast decision model like Jev cut per-decision cost by 66.1% to 78.4% and decision time by 73.9% to 81.5%, yet task success held near baseline on ALFWorld and fell more sharply in two mobile agents, so local savings do not automatically become delivered outcomes.

Across 5,601,890 Texas crashes from 2017 to 2025, a model reading the coded table and a model reading officer narratives were joined under one sampling design to produce population counts with intervals: narratives document 15,074 phone-use injury crashes against 7,340 in the coded field.

Across 5,601,890 Texas crashes from 2017 to 2025, a model reading the coded table and a model reading officer narratives were joined under one sampling design to produce population counts with intervals: narratives document 15,074 phone-use injury crashes against 7,340 in the coded field.