Free OffersModelsAPI PricingCoding Agents

TypeSafe AI Jev: Access, Pricing, and Structured Decisions

Jev is TypeSafe AI's structured decision model. See its current Vercel AI Gateway listing, published input price, suitable tasks, and limits.

Published: Last checked:

Some links on this page may be affiliate or promotional links. We may earn a commission if you sign up, at no extra cost to you. Our recommendations are based on the research described in our Editorial Policy.

TypeSafe AI Jev model overview
Analyzed Foundation Models (2)Full LLM Leaderboard→
Gemini 3 Flash
Google
37.1SCORE
DeepSeek-V4-Flash-0731
DeepSeek
46.1SCORE

Jev is TypeSafe AI’s structured decision model. Instead of returning a paragraph for an application to parse, it evaluates supplied state against typed questions and returns structured values. TypeSafe documents three question types: Choice (select from options), Score (score against a rubric), and Noul (assess whether a statement is true). Choice and Score also return probabilities and confidence. Those outputs can help software route or rank cases, but they remain model judgments rather than deterministic truth.

The current Vercel AI Gateway listing shows input pricing starting at $0.04 per million tokens and lists a 32K context. Its provider table lists TypeSafe AI and DigitalOcean. Vercel’s page no longer presents the earlier temporary free offer; that promotion ended September 25, 2026. The OpenRouter listing for Jev 1.13 currently shows $0.042 per million input tokens and $0 per million output tokens, with a 64K context. These gateways expose different model/provider details, so check the provider-specific terms and your account’s current billing view before sending requests.

What the model is suited to

TypeSafe describes Jev as a model for fast, structured decisions that application code can consume. Potential fits include:

  • Selecting a queue or department from a fixed set of options.
  • Scoring a case against a defined rubric.
  • Checking whether a statement appears true based on supplied context.

The docs advise keeping each question specific and well-scoped. If a decision requires balancing several independent factors or extended reasoning, ask smaller questions and combine their results in application code. That approach makes the policy visible and editable, while keeping each model question focused.

The probabilities and confidence values are useful signals, not guarantees of calibration or correctness in every domain. Test the model against representative cases, inspect failure modes, and set a human review path for decisions with meaningful consequences. Avoid describing it as “deterministic” or as a substitute for validating the underlying evidence.

Cost example

At the OpenRouter-listed rate of $0.042 per million input tokens, 100,000 calls averaging 500 input tokens each total 50 million input tokens and cost $2.10. OpenRouter lists output at $0/M for Jev 1.13. This is an arithmetic example, not a quote for other gateways: rates, usage metering, and any other charges can differ. Recheck the live rate for the provider you choose.

Example integration

Use TypeSafe’s Quick Start for current SDK setup and API syntax. Follow the documentation for the provider and SDK you select, since gateway endpoints and response shapes are provider-specific.

Verdict

Jev is worth evaluating when an application needs typed choices or rubric scores and can validate their quality against its own examples. The current Vercel listing is paid usage, not a guaranteed ongoing free tier. Start with the live provider terms and benchmark the model on the exact decisions your system will make.

Frequently Asked Questions

Is Jev free to use?

Do not assume ongoing free access. Vercel currently lists input pricing starting at $0.04 per million tokens; OpenRouter lists Jev 1.13 at $0.042/M input and $0/M output. The earlier Vercel free promotion ended September 25, 2026.

What does Jev do?

Jev evaluates supplied state against typed questions and returns structured choices, scores, or a truth-likelihood value with probabilities and confidence for some question types. These are model judgments, not guaranteed facts.

How much would 100,000 calls with 500 input tokens each cost at $0.042 per million?

That workload uses 50 million input tokens, or $2.10 at $0.042 per million on the OpenRouter listing, before any other fees or provider-specific terms.

Can Jev replace a generative language model?

Jev is designed for bounded structured decisions such as choosing, scoring, or checking a statement. Use a generative model when a task needs free-form explanation, synthesis, or extended reasoning.

Sources & disclosures

Source links for data and claims discussed in this article are listed below. Refer to each source for its current figures and methodology.

Share Article

Recommended Reading

Related by topic and recency.

All articles→
ReviewsModels

GLM-5.3-FlashX: 200 Tokens/sec Throughput vs 2.5x Price Analysis

Review GLM-5.3-FlashX provider speed claims, current API rates, task-cost math, and how to measure whether faster serving helps your coding workflow.

Sep 20, 2026 · 5 min read
DealsFireworks AI

Fireworks AI Credits: What New Developers Should Check

Fireworks AI billing and promotional credits can vary by account. Check the current offer and model pricing before estimating how much inference a balance covers.

Aug 12, 2026 · 2 min read
GuidesModels

Jev by TypeSafe AI: System One Decision Models Explained

How TypeSafe's Jev turns structured state into typed choices, scores, and probabilities, and where a specialized decision model may fit in software.

Sep 20, 2026 · 3 min read