Deepseek makes its 75 percent discount permanent, pricing output tokens at least 34x below GPT-5.5

Deepseek makes its 75 percent discount permanent, pricing output tokens at least 34x below GPT-5.5

Deepseek has made a significant move by keeping the 75 percent discount on its flagship model, Deepseek V4 Pro, permanently in place. The company announced this decision on their official platform.

The original promotion was set to expire on May 31, 2026, but now it will continue indefinitely at a reduced price. Under the permanent discount, one million input tokens without cache cost just $0.435, while one million output tokens cost $0.87. By comparison, GPT-5.5 charges $5 per million input tokens and $30 per million output tokens, while Opus 4.7 sits at $5 for input and $25 for output.

This pricing structure makes Deepseek’s flagship model about 11.5 times cheaper than GPT-5.5 on standard input pricing, and the gap widens significantly to around 34.5 times cheaper for output tokens when compared to GPT-5.5.

Token prices only tell half the story

The raw per-token pricing is just one part of the equation. The actual token consumption per task matters as much, similar to how the fuel efficiency of a car affects its overall cost. A low price per gallon doesn’t help if your engine guzzles it up.

For example, Google’s Gemini Flash 3.5 is cheaper and performs similarly to the previous Pro model 3.1, but it consumes far more tokens, potentially making it pricier in practice. Anthropic‘s Opus 4.7 looks cheaper on paper than GPT-5.5 too, but uses more tokens. Meanwhile, GPT-5.5 consumes fewer tokens than its predecessor.

Deepseek V4 clearly lags behind the top frontier models like GPT-5.5 and Opus 4.7 in raw performance. The actual cost depends on the task, and benchmarks only tell part of the story; real-world use will be crucial to determine how much cheaper Deepseek is.

As AI usage grows, companies are becoming more price-sensitive. Many firms may shift strategy from using the best model that still meets their needs to opting for a cheaper one that’s still good enough.

AI News Without the Hype – Curated by Humans

Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive “AI Radar” frontier report six times a year, full archive access, and access to our comment section.

Subscribe now

Key Takeaways

• Deepseek has made its flagship model’s discount permanent at 75% off.
• This pricing strategy makes Deepseek V4 Pro about 34 times cheaper for output tokens compared to GPT-5.5.
• The actual cost of using an AI model depends on how many tokens it consumes per task, not just the price per token.

Originally published at the-decoder.com. Curated by AI Maestro.

Stay ahead of AI. Get the most important stories delivered to your inbox — no spam, no noise.

Please enable JavaScript in your browser to complete this form.

Name Email

Name *

First

Last

Email *

Submit