BREAKING: DeepSeek Makes 75% Discount on V4-Pro API Permanent – The AI Price War Just Went Nuclear May 22, 2026 — In a move that’s sending shockwaves through the AI industry, Chinese AI lab DeepSeek has announced it is making its aggressive 75% discount on the flagship DeepSeek-V4-Pro model permanent.
@deepseek_aiWhat was originally a limited-time promotion (extended once until May 31, 2026) is now the new baseline pricing. This cements V4-Pro as one of the most cost-effective high-performance reasoning models on the market.New Permanent Pricing (effective after May 31, 2026):
- Input Tokens (Cache Miss): $0.435 per million (down from $1.74)
- Output Tokens: $0.87 per million (down from $3.48)
- Input Tokens (Cache Hit): $0.003625 per million (~99.8% savings on cached prompts)
Using standard blended pricing metrics, this puts V4-Pro at roughly ~$0.18 per million blended tokens — territory once reserved for much smaller, lighter models.
deepseek.aiV4-Pro is a massive 1.6T-parameter Mixture-of-Experts model (with ~49B active parameters) featuring 1M-token context, exceptional coding, math, and long-context reasoning capabilities. Why This Matters
- Developers & Builders Win Big: High-volume use cases such as agentic workflows, large-scale data processing, coding agents, RAG systems, and batch inference have just become dramatically more affordable. Many are calling it “AI for everyone.”
- Intensifies Global Price War: DeepSeek is undercutting Western frontier models (GPT-5.5, Claude Opus/Sonnet, Gemini) by 10-35x on output tokens while delivering competitive reasoning performance. Chinese labs continue aggressive pricing pressure.
- Cache Hits Are Insanely Cheap: The near-zero cache-hit pricing rewards smart prompt engineering and repeated context reuse — a huge boon for production apps.
Community Reactions are overwhelmingly positive:
- Builders are flooding timelines with usage screenshots showing millions of tokens for pennies.
- “China can have all my data” and “Welcome our new Chinese overlords” memes are trending.
- Comparisons to BYD in EVs and Android in mobile — “good enough + dramatically cheaper” wins volume.
- Some note limitations (e.g., native vision support), but for text-heavy reasoning workloads, it’s a game-changer.
This follows DeepSeek’s recent V4 series launch and positions the company as a leader in cost-effective frontier intelligence. Expect more adoption in open-source agents, startups, and enterprises optimising for ROI.DeepSeek’s official announcement: “We are making our discount permanent! 
Enjoy building with DeepSeek-V4-Pro and bring your innovative ideas to life! 
”The era of affordable, high-context AI reasoning is here. If you’re building anything at scale, now’s the time to test V4-Pro.What are you going to build with these new economics? Drop your thoughts below. 
#DeepSeek #V4Pro #AIPricing #AI #LLM
