deepseek-chat (legacy alias)

Legacy model-name alias, NOT a separate model. Official docs: 'deepseek-chat & deepseek-reasoner will be fully retired and inaccessible after Jul 24th, 2026, 15:59 (UTC Time). (Currently routing to deepseek-v4-flash non-thinking/thinking).' deepseek-chat = NON-thinking mode of deepseek-v4-flash. Exact retirement timestamp: 2026-07-24 15:59 UTC. Replacement: use explicit name deepseek-v4-flash (non-thinking). Pricing shown is the rate that was in force when this alias was retired, and it deliberately no longer tracks deepseek-v4-flash: DeepSeek moved the live API to peak/off-peak billing at 16:00 UTC on 2026-08-16, roughly three weeks AFTER this alias became inaccessible, so the new schedule never applied to it. Until 2026-08-16 this note read 'Pricing shown matches deepseek-v4-flash since it routes there', which stopped being true the moment the target row was re-pointed. This alias historically mapped to DeepSeek-V3-series non-thinking; as of the V4 Preview it routes to V4-Flash. Confirmed RETIRED on 2026-09-02. Signal: DeepSeek's own announcement is unconditional and timestamped — "deepseek-chat & deepseek-reasoner will be fully retired and inaccessible after Jul 24th, 2026, 15:59 (UTC Time)" — with no "earliest possible date" hedge, and that moment is 40 days past. Corroborated (not established) by absence: neither id appears on api-docs.deepseek.com/quick_start/pricing/ nor in the current list-models documentation, both re-read 2026-09-02. Note this is a forward-looking declaration plus absence, NOT a past-tense confirmation of the kind OpenAI publishes; DeepSeek maintains no live status table to re-read. Until 2026-09-02 the row was held at "deprecated" because the date alone was not evidence.

Source: DeepSeek official documentation ↗

Explore more

deepseek-chat (legacy alias) — questions & answers