DeepSeek V4.1 Flash 1017: what DeepSeek says changed, and swapping it in without breaking anyone's saved model

Описание и применение

Миграция модели DeepSeek V4.1 Flash 1017 в LU Cloud была выполнена без разрыва существующих пользовательских сценариев: старый идентификатор перенаправляется на новый, сохраняется совместимость и снижаются расходы.

Задача

Users had V4.1 Flash set as default in chats and API scripts; renaming the model would break those saved picks.

Результат

The migration preserved all existing user workflows, avoided breaking saved picks, and lowered cloud costs from 30,000/120,000 to 20,000/60,000 credits per million tokens.

Рабочий процесс

• 1. DeepSeek released V4.1 Flash 1017 with higher benchmark scores and lower pricing.
• 2. Locally Uncensored updated its catalog to add the new id and a retired‑ids table mapping the old id to the new one.
• 3. The catalog lookup ensures all requests to the old id are routed to the new model and priced accordingly.
• 4. Users can continue using the old id in chats and API scripts without changes.
• 5. The desktop app automatically pulls the updated catalog; code clients use the new id with the LU Cloud endpoint.

Доказательства

People had V4.1 Flash saved as a default, in old chats and in API scripts that send deepseek-ai/DeepSeek-V4.1-Flash.

Источник

The catalog entry now carries the new id, deepseek-ai/DeepSeek-V4.1-Flash-1017, and the old id sits in a small table of retired ids that points to its replacement.

Источник

1017 draws 20,000 credits per million input tokens and 60,000 per million output tokens. Before the switch the catalog listed V4.1 Flash at 30,000 and 120,000.

Источник