DeepSeek API adds v4-pro and v4-flash with 1M context and legacy endpoint retirement

teortaxesTex · x · 2026-07-22

DeepSeek’s API now supports deepseek-v4-pro and deepseek-v4-flash while keeping the same base URL. The models expose both OpenAI ChatCompletions and Anthropic-compatible APIs, and each supports 1M context plus two modes: Thinking and Non-Thinking.

The quote also highlights the planned retirement of legacy deepseek-chat and deepseek-reasoner endpoints on Jul 24, 2026, 15:59 UTC. Pricing shown in the screenshot puts v4-pro at $0.145 / $1.74 / $3.48 for input cache hit, input cache miss, and output respectively, while v4-flash is much cheaper at $0.028 / $0.14 / $0.28.

Original post →

More from Infra

Infra channel →