backend new models check - deepseek - 2026-09-16

Files checked: `src/backend/drivers/ai-chat/providers/deepseek/models.ts`

**chat (1)**
- `deepseek-flash` | aliases: none upstream | context: n/a | created: n/a | price: see provider pricing page
Notes: Upstream returns only `{id, object, owned_by}` — no aliases/context/created/pricing fields at all. "deepseek-flash" does not match any backend id/alias (backend has `deepseek-v4-flash`, `deepseek-chat`, etc., which are distinct strings), so it's reported new per the literal name-match rule, but it may simply be an upstream rename/short-alias of the already-supported `deepseek-v4-flash` — worth a human check against DeepSeek's docs before adding as a wholly new model vs. an alias.

skipped (unsupported modality): 0
legacy upstream models not in backend (verify intent): none

Start with src/backend/drivers/ai-chat/providers/deepseek/models.ts, then check DeepSeek's documentation to determine whether deepseek-flash is a new model or an alias for deepseek-v4-flash. Done means the model is represented correctly in the backend according to that verification, with the existing model naming and alias rules preserved.

Written by the indexing model from the issue text.