DeepSeek made its name as the price butcher of AI APIs. That ends August 17. Alongside the launch of V4 Pro — a 1.6T-parameter MoE with a 1M-token context window — DeepSeek repriced its whole API line: V4 Flash goes from $0.14 to $0.27 per million tokens (+93%), V4 Pro peak output hits $3.96/M versus $0.87 before, and cache-miss input jumps to $1.32/M. Depending on model, token type, and hour, some workloads now cost up to 1,100% more.
What changes for API users
This is an API story: same endpoints, same models, new bill. DeepSeek is adding peak/off-peak pricing — off-peak output runs $1.98/M, half the peak rate. If you run coding agents or batch pipelines on DeepSeek, the move is obvious: push heavy jobs into off-peak windows. V4 Pro itself targets exactly those agents, with adjustable reasoning effort and an 87.9 on Terminal-Bench 2.1.
The real signal
The same week, OpenAI cut GPT-5.6 Luna prices by 80%. The cheapest player raising prices while the incumbent slashes them is the clearest sign yet that the subsidized-API era is closing. DeepSeek is still cheaper than most Western rivals. But the direction just flipped.
You Might Also Like
- Deepseek v4 pro Hits gpt 5 Parity on 5 of 7 Benchmarks at a Fraction of the Cost
- Deepclaude Lets Claude Code run on Deepseek v4 pro 0 87 vs 15 per Million Tokens
- Openai gpt Realtime 2 Translate Whisper Three Voice Models one api Several Startups Erased
- Openai gpt 5 6 sol Terra Luna a Three Tier Lineup Only 20 Orgs can Touch
- Deepseek v4 pro v4 Flash Ship With 1m Context and 0 28 m Output
Top AI Product 
Leave a comment