DeepSeek API Consolidates to V4 Models, Sunsetting Legacy Chat and Reasoner Endpoints

Article body

Developers using the DeepSeek API must update their model references before July 24, 2026, when deepseek-chat and deepseek-reasoner are officially deprecated. The deprecation was flagged in the company's official API documentation, which notes that deepseek-chat maps to the non-thinking mode of deepseek-v4-flash, while deepseek-reasoner maps to the thinking mode of the same base model. The retirement is part of a broader consolidation of DeepSeek's model portfolio around its newer V4 generation. The DeepSeek API maintains compatibility with both OpenAI and Anthropic SDKs, using a base URL of https://api.deepseek.com for OpenAI-format calls and https://api.deepseek.com/anthropic for Anthropic-format calls. Available models currently include deepseek-v4-flash, deepseek-v4-pro, and the two legacy identifiers pending retirement.

Why this matters

• Any application currently calling deepseek-chat or deepseek-reasoner will break after July 24, 2026, requiring code changes before that date.
• The consolidation means developers need to explicitly choose between the fast, non-thinking deepseek-v4-flash and the more capable deepseek-v4-pro for their use cases.
• Because DeepSeek's API format mirrors OpenAI and Anthropic conventions, migration is straightforward — the primary change is swapping model identifier strings.

Source note

• Information drawn from DeepSeek's official API documentation at api-docs.deepseek.com, which specifies the deprecation timeline, base URLs, and current model lineup. Developers should monitor the same documentation for any updates to the July 24, 2026 cutoff date.

Original link

Open the monitored source