Jev Article

TypeSafe's Series A announcement: $870M at a $7.5B valuation led by Andreessen Horowitz with Sequoia and DCVC, and what the post says the money changes for developers, businesses, and recruits.

A TypeSafe case study of a talent marketplace that swapped Gemini 3.1 Flash Lite for Jev in candidate matching: 88% lower cost, about twice as fast, and a projected $500K a year saved.

Nikola Balić on why the Jev launch was easy to try and share: a side-by-side demo, a first useful decision, and distribution through people who built their own demos.

swyx's Latent Space interview with Diogo Almeida, posted on Hacker News as the first long technical conversation with TypeSafe's CEO.

InfoQ's 1 October 2026 news brief on Jev: a parallel pass over typed questions, TypeSafe's published price and latency, and where to read the jaggedness notes.

Tim Fernholz's 18 September 2026 TechCrunch piece on Jev: no text output, synthetic-data training, and a launch that briefly outran the API.

Philipp Acsany's tutorial replaces a Y-or-N input check with Jev, then asks Noul, Score, and Choice in one call through OpenRouter.

Tom Farnschläder's 24 September 2026 tutorial builds a support-ticket router with the Python SDK, using Choice, Score, and Noul in one call.

Kenny Rogers explains Jev as a decision model and shows Choice, Score, and Noul responses from calls he placed through OpenRouter on 21 September 2026.

Yury Selivanov's 2 October 2026 post on the Python AI SDK: experimental evaluate() calls typesafe-ai/jev through an AI Gateway key.

Databricks' 30 September 2026 announcement of ai_decide, a beta SQL and REST function for typed decisions, described as compatible with the TypeSafe API.

Together AI's tutorial for fine-tuning a Jev-like classifier, Tev1, on Qwen3.5 4B and serving it on Together. The example uses 38,340 rows.

A preregistered Quick, Draw! study: 400 SVG-sketch classifications with Jev versus Sonnet on the same text, plus base64 and image baselines.

Requesty's September 2026 guide to Jev as a System One model: state, Choice, Score, and Noul, plus how Requesty routes typesafe/jev-1.13.0.

Powerdrill Bloom on what shipped with Jev 1.13.0: RLCD, the three question types, pricing, and TypeSafe's jaggedness page.

OmniaKey's fact-checked table for jev-1.13.0, limits, and pricing, plus a live connectivity check through OmniaKey's alpha search endpoint.

Kie.ai on Jev as a System One evaluation model: parallel typed questions, pricing, and where TypeSafe says hosted weights are not confirmed yet.

Vercel's changelog for typesafe-ai/jev on AI Gateway, with AI SDK 7 experimental_evaluate and a support-ticket routing example.

Opper hosts Kev 4B and typesafe/jev-latest on the same /v1/systemone compat route and compares accuracy, tokens, and latency on 362 post-release questions.

James Cruce replays held-out routing decisions with local Laya against deterministic ordering, not a full Laya-versus-Jev accuracy bake-off.

Datadog's walkthrough of one Jev rubric that scores every criterion in a single request, then runs as online evals on live spans and as offline experiments.

Vercel's guide for when a typed Jev decision is enough and when a job still needs GPT-6 Astra to write, see images, or call tools.

Sam Reghenzi's product-taxonomy benchmark: a Jev Choice at every level versus a gpt-5.2 agent that descends the tree and can be sent back by a judge.

Simon Willison's overview of System One: one state, many parallel questions, and Noul, Choice, and Score instead of generated text.

Alex Molas argues that a probability fit on TypeSafe's data is not calibrated on yours, so Jev's numbers should be treated as scores.

Allan Riordan Boll reads one-token logprobs so a chat model returns a letter and a distribution, including on webcam frames from a vision model.

Privatemode's write-up of typed decisions from GLM-5.3-Flash in one forward pass, with latency, cost, and accuracy set next to Jev and Laya.

Kunal Singh records coding-agent actions with Gryph and asks Jev yes-or-no questions against a short profile of how that developer usually works.

Vercel's guide for seven jobs Jev can take: routing forms, prioritizing tickets, reviewing tool calls, picking a model, categorizing documents, moderation, and checking an answer.

Rodney L. trains OpenJev to play cooperative Overcooked. Two copies of one policy serve six soups in 512 ticks, and every action is a single native button.

OpenRouter's Banking77 run: 3,080 support utterances, 77 intents, Jev 1.13 against Claude Opus 5 on the same one-line criteria, with latency, cost, and a confidence cascade.

OpenRouter times Jev 1.13, GPT Luna, and Claude Opus on 60 support tickets: intent plus an escalation flag, then says the reply still belongs to a generative model.

Five self-funded experiments on jev-1.13.0 from a Mac mini, comparing direct HTTP with claude-sonnet-5 after an earlier test mistook CLI startup for model speed.

A black-box pass over ten public datasets with jev-1.13.0: injection detection, reranking, intent, tool routing, and three tasks the author says do not work.

Archestra labels 100 production Claude Code tool calls and compares Jev with a constant benign baseline, Sonnet 5, Laya, and several open decoders.

Tom Nick's write-up before jevdit.com: irony detection on 100 German soccer tweets, with Jev 1.13.0 next to Opus, Astra, and Laya.

A parody that classifies one email by reading A/B/C logits from a local Qwen3-0.6B GGUF, then says this shape is what people mean by Jev.

John Berryman's argument that OpenAI already uses its models as implicit classifiers and could ship a general one inside agents, if it can match the training.

An explainer written for people who read everything about Jev and still did not get it, published as an X article and shared as a post.

A release guide with eval tables, pricing, curl and Python examples, and fit boundaries for Jev's early access.

A heavily footnoted claims audit that separates documented Jev facts from marketing, including the seed round, third-party tests, and missing transparency.

A hands-on Python guide that installs typesafe-sdk, walks through Choice, Noul, and Score, and sketches a chargeback-tool integration and benchmark plan.

A TypeScript tutorial and critical explainer covering AI SDK integration, triage code, and the zero-hallucination caveat.

A Japanese experiment that builds a rules-based gomoku harness and replays a full Jev-versus-Jev game log to measure real per-move latency.

A skeptical Japanese replication that approximates Jev's speed by reading first-token logits from small LLMs, with Gemma3 and a Mario benchmark.

TypeSafe's case for machine-native composable AI: intelligence as a dependable primitive that software branches on, rather than an assistant that keeps humans in the loop.

A TypeSafe blog essay arguing that in ML the order that matters is doing the right task, then data, then compute, then algorithms, using the InstructGPT result as its example.

Wikipedia's article on Jev: TypeSafe AI, the September 15, 2026 early-access release of jev-1.13.0, the Choice, Score, and Noul primitives, RLCD training, and the unpublished weights and technical paper.

The Register's launch report on TypeSafe AI's Jev: the $40 million raise, typed probabilistic decisions, the Doom demo, the 70-500 ms latency and $0.042/MTok pricing claims, and the structured-output caveat.

Tom's Hardware's launch report on Jev, covering how the API takes program state and typed statements and returns yes/no answers, choices, or probability distributions with confidence values.

Latent Space's AINews roundup of launch day: the Jev thread atop Hacker News, the RLCD decision-model framing, and community readings that compare it to DSPy-style typed signatures for routing and scoring.

News analysis arguing Jev's refusal to generate text is the feature, covering the $40M seed, the OpenAI-veteran founding team, and what structured, thresholdable decisions mean for agent pipelines.

DataCamp's explainer walks the eval tables behind System One models, from 67.8% workflow accuracy and 0% structured-output errors to the $0.042 per million token price and how to call the API.

A use-case study comparing Jev, Mistral Small 4, and Gemini 3.5 Flash-Lite on rejecting unsuitable local-event listings, with individually tuned prompts, accuracy, response times, and costs.

A 2026 technical SEO audit price survey plus a measured case where a Jev pipeline crawled and judged 1,204 pages, producing 4,816 typed judgments in under three minutes for under $1 in model cost.