Hamed Nilforoshan evaluated OpenAI Decisions API against Jev on HiringCafe, reporting that OpenAI's API was twice as expensive and 5% to 10% worse for ranking job query relevance.
Modal reported that TypeSafe AI scaled its Jev deployment on Modal infrastructure, serving one trillion tokens between the model's Thursday launch and Sunday.
Google Gemma says seeding a canvas with a response template in vLLM lets DiffusionGemma return confidence scores and probability distributions for yes/no, multiple-choice, and scored questions in a single denoising step.
Zhaorun Chen says red-teaming Jev 1.13 on DTap found a 70.1% attack success rate for direct misuse and 43.5% for indirect prompt injection. He says using Jev to gate its own tool calls reduced that rate while keeping most of its utility; TypeSafe AI questions whether Jev is failing and says it is built for composability.
On JevBench, OpenRouter says solar-decide scored 87.0% accuracy versus 86.1% for Jev 1.13, with a median latency of 0.14 seconds versus 0.30 seconds.
Diogo Almeida spoke with a16z’s Ben Horowitz and Martin Casado about Jev, a model built to live inside software. It reads natural language and returns one choice from a set of options, with a confidence level for each.
TypeSafe AI says new Jev signups no longer receive free credits and hopes to restore them as soon as possible. Jev can be started in the TypeSafe console.
A paper reports that TypeSafe AI’s Jev, with no extra training, separates alignment failures from good responses at a median AUROC of 0.886. Across 19 benchmarks, one Jev pass cost $0.30, compared with $18.96 for the LLM judges those benchmarks use.
OpenRouter says Jev accounted for 27% of weekly classification request volume, nearly twice the share of DeepSeek V4 Flash, which previously led that category.
On a small support agent, elvis says Jev Router and a fixed GPT-6 Sol baseline both solved all 8 cases across 32 calls. The router cost $0.008 versus $0.018, with a 1.5-second median response time versus 1.9 seconds.
Isaac Miller says DSPy 3.4.0 adds native support for Jev and System One models and introduces ReAnchor, an optimizer for calibrating outputs with confidence.
OpenRouter says developers already calling Jev can switch to Kev by changing only the model field to jaredpalmer/kev-4b.
A Jev-as-a-Judge paper reports that accepting TypeSafe AI’s confident JEV verdicts and escalating the rest to GPT-6 Astra kept 99% of GPT-6 accuracy on 510 held-out preference pairs, at about 57% of GPT-6’s fee. The authors say the escalation threshold did not transfer to every fallback model and should be set on one’s own data.
Pydantic says the agents use TypeSafe AI’s Jev classifier to answer typed questions, returning each agent’s existing output model with one confidence score per field.
On September 22, 2026, Ahmad Awais said he was launching Jev on Command Code for free, limited to headless mode or the Provider API. He also open-sourced the mod cmd-mod-jev-nudge.
Jev can be used only in headless mode or through the Provider API. Ahmad Awais also open-sourced the mod cmd-mod-jev-nudge and says the free access is for keeping an agent running while more work remains.
Shahriar Tajbakhsh said Metaview added TypeSafe AI's jev to every agent over the weekend. He said candidate searches in its sourcing product now take seconds instead of minutes, with the same accuracy, about 10 times the speed, and a meaningfully lower cost per search.
LLM requests can be classified automatically with typesafeai's Jev in OpenRouter Classifiers, and the results can be analyzed in Explore.
TypeSafe quoted MotherDuck’s claim that prompt_jev(), a SQL function powered by TypeSafe’s Jev model, made text classification about 50x faster at about 1% of the cost. MotherDuck reported 100k rows in 40 seconds for $0.50 versus 32 minutes and $37 for an LLM.
TypeSafe AI said Jev is now available and linked to https://console.typesafe.ai. The post did not specify what Jev is or any access terms.
