DeepSeek-V4.1-Flash now runs the 1017 release

deepseek-v4-1-flash on cheapaimodel is now answered by DeepSeek-V4.1-Flash-1017, the post-training update DeepSeek released on 17 October 2026. It replaced the September release under the same model ID, so requests that send deepseek-v4-1-flash reached the new release without any change on your side.

The switch took effect on 2 November 2026. If your outputs changed around that date, this is the likely reason.

What changed

According to DeepSeek, the 1017 release scores better on its benchmarks, finishes tasks faster and writes shorter outputs than the September release. Run your own evaluation on your prompts before you rely on that.

Nothing else changed:

• Price: unchanged. See DeepSeek-V4.1-Flash for current peak and off-peak prices.
• Limits: 1M-token context window, up to 384K output tokens.
• Model ID: keep sending deepseek-v4-1-flash.

Also new on 2 November

• The model field of responses now reads deepseek-v4.1-flash-1017. It differs from the name you send; both refer to the same model.
• Two aliases were added: deepseek-v4-1-flash-1017 and deepseek-v4.1-flash-1017. They reach the same release and are billed the same way. If your API key has an allowed-models list, add them to it before you send them.
• The model page has a version section listing the current release, the name in responses and the accepted aliases.

The September release is no longer available through cheapaimodel. When DeepSeek ships its next update, deepseek-v4-1-flash will follow it and we will post it here.

DeepSeek-V4.1-Flash-1017: what changed and how to call it has code samples and the full price table.