Overview
DeepSeek Harness (DSH) 推理等级自动选择插件：由 Jev System One 模型判断每条消息值多少思考量，按模型声明的等级自动推导档位，上下文信封让「继续」这类追问继承话题深度，低置信度向上取，任何失败都静默沿用原等级。 | Jev-driven reasoning effort per message: per-model ladders derived from what each model advertises, a fixed-size context envelope so follow-ups inherit topic depth, ties break upward, silent fallback on every failure path.
Pricing
Free
Detailed plans have not been confirmed in our catalog. Check the official website for current limits and billing terms.
Visit WebsitePrices and limits may change. Confirm the currency, billing period, seat minimum and usage caps on the official website.
Use Cases
- Teams use the plugin to avoid wasting high reasoning effort on simple messages like greetings.
- Teams use the plugin to automatically raise reasoning effort for complex refactoring tasks.
- Teams use the plugin to let Jev decide the reasoning level for each message based on its content.
- Teams use the plugin to keep the same reasoning level across all steps and retries within a turn.
- Teams use the plugin to respect manual effort level or model changes made in the selector.
- Teams use the plugin to disable Jev for a specific session without affecting other sessions.
Features
Selects reasoning effort per message
Uses Jev to assess thinking needs
Derives levels from model capabilities
Preserves topic depth across follow-ups
Uses a fixed-size context envelope
Rounds uncertain choices upward
Retains original level on failure
Reviews
No reviews yet. Be the first to share your experience!