JevTypeSafe
Try the Jev and Laya judgment models in your browser to classify, make yes/no judgments, or score text, then reuse the same configuration via API, CLI, and MCP.
About JevTypeSafe
JevTypeSafe lets you try the Jev and Laya System One judgment models in your browser: request a category, a yes/no judgment, or a score, view the returned calibrated results, and reuse the configuration in your own development workflow. Jev is the flagship model, supporting rich-text documents and questions of up to 8,000 tokens; Laya English (512 tokens) and Laya Multilingual (1,024 tokens) are ultra-fast, lightweight models designed for high-throughput short text, comment moderation, and real-time LLM guardrails, with latency under 0.35 seconds. The Playground and API share the same account quota, and the same judgments can also be called from applications, terminals, or agents via the HTTP API, CLI, or local MCP.
Best for
Best use cases
Classify, make yes/no judgments, or score long document content
Moderate short texts such as comments in batches
Add real-time guardrail judgments to LLM applications
Integrate judgments into your development workflow via API, CLI, or local MCP
Features
Jev: flagship System One model
Supports rich-text documents and questions of up to 8,000 tokens, ideal for judgment scenarios that need full context.
Laya: ultra-fast lightweight models
Laya English (512 tokens) and Laya Multilingual (1,024 tokens) are designed for high-throughput short text, comment moderation, and real-time LLM guardrails, with latency under 0.35 seconds.
Multiple input options
After choosing a model, send text or JSON input—up to 16 questions can be answered in a single request, with categories, yes/no judgments, or scores as needed.
API, CLI, and local MCP
The same judgments can be called from an application, the command line (npx jevtypesafe), or a local MCP interface, with account quota shared across all entry points.
Pros
- Output tokens are free; you only pay for input tokens
- Requests that do not return a valid judgment result are not billed
- Get 1,000,000 input tokens on sign-up, shared between the Playground and the API
- Unused tokens carry over, and token packs never expire
- Multiple integration options: HTTP API, CLI, and local MCP
Pricing Plans
Creator
- 720M input tokens / year
- Billed by input tokens
- Output tokens free
- Unused tokens carry over
- Cancel renewal anytime
Studio
- Input token billing
- Output tokens free
- Unused tokens roll over
Max
- 4.752B input tokens / year
- Includes an extra 20% tokens
- Billed by input tokens
- Output tokens free
- Unused tokens carry over
* Pricing information is for reference only. Please refer to the latest announcement on JevTypeSafe's official website.