Jev(latest) API pricing and integration guide
Review Jev System One decision routing, documented 64K context, Choice/Score/Noul questions, and API integration boundaries.
Current prices and model reference
- Model ID
jev-latest- Documented context
- 64K
- Documented image input
- —
- Live model multiplier
- Expression billing
- Input price
- —
- Output price
- —
Prices are in CNY. A 1× multiplier equals CNY 0.4 per million input tokens. Output and cache charges follow the billing configuration; your usage record is authoritative.
View the full live pricing catalogWhen to consider this model
Consider it for ticket routing, risk gates, relevance scoring, agent routing, and human-review triggers. It returns typed decisions, not ordinary generated text.
How to evaluate it
Start with labeled classification or scoring samples. Record probabilities, confidence, human-review rate, and final actions before deciding whether to pin a version ID.
What to check before integration
The new Feishu table lists 2x, CNY 0.8 per million input tokens, with no output or cache charge. The live route uses expression billing; its generic ratio field is not an effective rate. Use jev-latest with POST https://api.llm-token.cn/v1/decisions (or /v1/systemone), state, questions and model, not /v1/chat/completions or /v1/messages.
Client configuration
In a client supporting a custom API endpoint, enter this base URL and model ID with your own API key. Select the protocol described in the client tutorial and validate a short request first.
Base URL: https://api.llm-token.cn/v1
Model ID: jev-latest