Are deepseek-v4-flash and deepseek-v4-flash-vision-exp still working?

DeepSeek still accepts those strings temporarily, but the underlying V4 Flash models are retired. The pricing page footnote says legacy names are served by DeepSeek-V4.1-Flash-1017 and billed at Flash prices. DeepSeek does not publish an end date for the temporary routing.

What about deepseek-chat and deepseek-reasoner?

DeepSeek announced on 2026-04-24 that both names would be discontinued on 2026-07-24. Before discontinuation they pointed at deepseek-v4-flash non-thinking and thinking modes. They are not listed on the current Models & Pricing table. We did not verify whether the API still accepts them after 2026-07-24.

Is deepseek-v4-pro still the right Pro name?

Yes — deepseek-v4-pro remains the documented API name for V4 Pro. The pricing table lists deepseek-v4-pro alongside deepseek-flash. The 2026-09-10 change log confirms continued V4 Pro API service after 14 Sep 2026.

DeepSeek API model names: what to use now

For the current Flash line, call deepseek-flash. DeepSeek documents this as the API name for DeepSeek-V4.1-Flash (including the 2026-10-17 DeepSeek-V4.1-Flash-1017 refresh).

For V4 Pro, call deepseek-v4-pro. The 2026-09-10 change log says V4 Pro API service continues after 14 Sep 2026 with billing unchanged until DeepSeek gives further notice.

Legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp are retired but still accepted temporarily; requests are routed to the current Flash model and billed at Flash prices per the pricing page.

Checked on 26 Oct 2026 against DeepSeek API change log.

Official sources

Current model names

API model name	Product	Status
deepseek-flash	DeepSeek-V4.1-Flash-1017	Current Flash name
deepseek-v4-pro	DeepSeek-V4-Pro-0813	Current Pro name
deepseek-v4-flash	Retired V4 Flash	Temporary alias → Flash
deepseek-v4-flash-vision-exp	Retired vision exp	Temporary alias → Flash
deepseek-chat	Legacy chat alias	Discontinued 2026-07-24 (announced)
deepseek-reasoner	Legacy reasoner alias	Discontinued 2026-07-24 (announced)

Product labels match the Models & Pricing table as of the checked date.

Retired and redirected names

deepseek-v4-flash and deepseek-v4-flash-vision-exp — retired 2026-09-10; still accepted temporarily and routed to the current Flash model (pricing footnote).
deepseek-chat and deepseek-reasoner — scheduled discontinuation 2026-07-24 (2026-04-24 change log). Not listed on the current pricing table.

Timeline of renames

2026-04-24 — V4 API names deepseek-v4-pro / deepseek-v4-flash; legacy chat/reasoner discontinuation announced.
2026-09-10 — Flash canonical name becomes deepseek-flash; old V4 Flash strings temporarily routed.
2026-10-17 — Flash backend refresh ( DeepSeek-V4.1-Flash-1017); still call deepseek-flash.

How to update your code

OpenAI-compatible SDK example:

Before → after (Python OpenAI SDK)

# Before (legacy Flash string)
client.chat.completions.create(model="deepseek-v4-flash", messages=[...])

# After (current Flash name)
client.chat.completions.create(model="deepseek-flash", messages=[...])

Anthropic-format clients: keep your existing base_url and change only the model field to the current name. dsh profiles: update the model string in your profile config to deepseek-flash or deepseek-v4-pro — see Install dsh.

How to check what you are calling

Log the model field on outbound API requests, grep your repo for retired strings, and compare integration tests against the official model table.

FAQ

Use deepseek-flash.

The 2026-09-10 and 2026-10-17 API change log entries say the API calling method and pricing stay the same while deepseek-flash serves the latest V4.1 Flash build.

DeepSeek still accepts those strings temporarily, but the underlying V4 Flash models are retired.

DeepSeek announced on 2026-04-24 that both names would be discontinued on 2026-07-24.

Yes — deepseek-v4-pro remains the documented API name for V4 Pro.