Review
After Jev landed, open-weight and hosted clones showed up fast. Laya Studio is one of the clearer hosted options for builders who want compatible requests without TypeSafe specifically.
Laya Studio hosts the open-source Laya decision model behind a Jev-compatible System One API, with optional MCP tools and a Guard endpoint for agent gates.
What that means in practice:
Last verified: October 2026 against laya.studio pricing/llms.txt signals and our Laya tool page.
| Path | Signal (Oct 2026) |
|---|---|
| Pay-as-you-go | ~$0.0357 / 1M input (1 credit = 1 input token) |
| vs Jev list | Marketed ~15% below ~$0.042/M |
| Monthly plans | From ~$19 with volume bonuses (Scale cited near ~$0.031/M) |
| Swiss-only residency | Higher credit rate (~$0.041/M cited) |
| Free starter | Small token allowance for new workspaces (confirm dashboard) |
Failed requests are marketed as not billed. Still read the live pricing page before you commit.
Auth pattern: Bearer lsk_live_… keys.
Why agents people care:
If you only need raw System One HTTP, you can ignore MCP. If you are wiring judgments into an IDE agent, MCP is the differentiator versus a bare model host.
Best for:
Not ideal for:
See also: what is a System One decision model, Jev vs Laya vs Clef.
Laya Studio hosts the open-source Laya System One decision model behind a Jev-compatible API. You send state and typed questions, get answers with probabilities, and can optionally use MCP tools and a Guard endpoint.
Pay-as-you-go is about $0.0357 per 1M input tokens on Laya Studio pricing materials, marketed as roughly 15% below Jev’s $0.042/M list. Monthly plans start around $19 with volume bonuses. Confirm laya.studio/pricing.
Laya markets a Jev-compatible wire format at POST https://api.laya.studio/v1/systemone. That makes swaps easier, but you should still regression-test questions and thresholds.
The Laya model lineage is open source (Apache-2.0 checkpoints cited from Convai Innovations). Laya Studio is an independent hosted service, not TypeSafe.
Laya Guard is an endpoint for tool-call style allow/ask/block verdicts, aimed at agent workflows. Pair it with the decision API when you need a hot-path gate.
3 curated tools below.
Laya Studio hosts the open-source Laya System One decision model behind a Jev-compatible API: send state and typed questions, get answers with probabilities in one forward pass, plus optional MCP tools and a guard endpoint.

Jev is TypeSafe's flagship System One decision model: send application state plus typed Choice, Score, and Noul questions, and get calibrated probabilities your code can branch on, without generating free-form text.

Cloudflare Clef and Clef-flash are open-source (Apache 2.0) System One decision models on Workers AI: multimodal state plus typed noul/choice/score questions return probabilities, with a Jev-compatible API shape and optional fine-tuning path.
| Tool | Best for | Pricing | Billing note |
|---|---|---|---|
| Laya | AI Agent | Freemium | Free Trial |
| Jev | Productivity | Paid | Paid Service |
| Cloudflare Clef | AI Agent | Freemium | Free Trial |
What is Laya?
Laya Studio hosts the open-source Laya System One decision model behind a Jev-compatible API. You send state and typed questions, get answers with probabilities, and can optionally use MCP tools and a Guard endpoint.
How much does Laya cost?
Pay-as-you-go is about $0.0357 per 1M input tokens on Laya Studio pricing materials, marketed as roughly 15% below Jev’s $0.042/M list. Monthly plans start around $19 with volume bonuses. Confirm laya.studio/pricing.
Is Laya compatible with Jev?
Laya markets a Jev-compatible wire format at POST https://api.laya.studio/v1/systemone. That makes swaps easier, but you should still regression-test questions and thresholds.
Is Laya open source?
The Laya model lineage is open source (Apache-2.0 checkpoints cited from Convai Innovations). Laya Studio is an independent hosted service, not TypeSafe.
What is Laya Guard?
Laya Guard is an endpoint for tool-call style allow/ask/block verdicts, aimed at agent workflows. Pair it with the decision API when you need a hot-path gate.