---
title: "What is Clef: the highest-priced of the three Clef text-and-image decision models | system-one.dev"
description: "Clef is the highest-priced of the three Clef text-and-image decision models on system-one.dev. When Jev or Clef Flash is not good enough, compare Clef on the same data. Sign in to try live."
canonical: "https://system-one.dev/en/models/clef"
language: "en"
---

# Clef

The highest-priced of the three Clef models. It reads text, JSON, and images and returns a choice, a score, or a yes/no answer.

**What is Clef?** Clef is a text-and-image decision model provided by system-one.dev. Give it text, JSON, or images and a few questions (pick one, score, or yes/no); it returns a choice, a score, or a yes/no probability and does not generate text. Among the three Clef models it has the highest rate. When Jev or [Clef Flash](https://system-one.dev/en/models/clef-flash) is not good enough, compare Clef on the same data. Call it with the TypeScript SDK or the API. Billing is in credits; the rate is in Model info on this page.

Sign in to try live: [Try live (sign in required)](https://system-one.dev/en/demo)


## Model info

- Provided by: system-one.dev
- 6 credits per question per 4KB
- Up to 32 questions · 13MB per request
- Inputs: text, JSON, and images (PNG, JPEG, WebP)
- Images: up to 4 · 4MB each · 8MB total
- Pick one: up to 255 options; Score: 2–10 levels
- Released 2026-10-01

## Uses

- **Classification comparison**: When Jev or Clef Flash is not good enough, compare Clef on the same data. See [Agent tool routing](https://system-one.dev/en/blog/agent-tool-routing).
- **Scoring**: Use Score against a rubric on an answer or a text-and-image result, as a first-pass evaluation. See [Using Score instead of an LLM as judge](https://system-one.dev/en/blog/llm-as-a-judge-alternative).
- **Guardrails**: Judge whether text and image content breaks a rule, then allow, block, or hand off from a yes/no probability. See [Agent guardrails](https://system-one.dev/en/blog/agent-guardrails).
- **Agent completion checks**: Look at a result or a screenshot and decide whether the task is actually done. See [When an agent should stop](https://system-one.dev/en/blog/agent-stop-condition).

## Frequently asked questions

### How is Clef different from Clef Flash?

Clef Flash, Clef, and Clef Omni share the same request format and all accept text and images. Clef Flash is the cheapest and fits high-volume, cost-sensitive work. Clef has the highest rate; when Jev or Clef Flash is not good enough, compare Clef on the same data. Compare the rates in each page’s Model info. See [What is Clef Flash](https://system-one.dev/en/models/clef-flash).

### Is Clef a large language model?

No. It does not write a reply or explain its reason. It only returns a choice, a score, or a probability. Use a generative model when you need written content.

### Does Clef support Chinese?

Test Chinese results with your own data first.

### How is it billed? Is there a free allowance?

By credits, from the number of questions and the request size. See Model info on this page. Details: [Pricing](https://system-one.dev/en/pricing). New accounts get 100 free credits on sign-up.

## Official sources

- [On this site: decision primitives](https://docs.system-one.dev/en/docs/primitives)
- [GET /v1/models](https://system-one.dev/v1/models)
- [All models](https://system-one.dev/en/models)

Last updated 2026-10-12

- [Open GET /v1/models](https://system-one.dev/v1/models)

- [Developer docs](https://docs.system-one.dev/en/docs/api#developer-details)

- [Try live (sign in required)](https://system-one.dev/en/demo)

- [All models](https://system-one.dev/en/models)