DeepSeek has announced an upcoming price hike for its API services. The company expects a substantial increase in rates, warning developers to adjust their usage accordingly. Specific pricing details will be shared in an official notice soon.
The news came via an official announcement from DeepSeek, which was first reported by IT Home. Developers currently rely on the platform for its extremely low rates. The looming adjustment represents a major shift for teams who built their software around the current low cost structure.
Right now, the pricing structure separates input and output tokens while accounting for cache status. The DeepSeek V4 Flash model, favored for high speed tasks, costs only 0.02 RMB per million tokens when a cache hit occurs. If the cache misses, that input price jumps to 1 RMB per million tokens. Output tokens for this model are priced at 2 RMB per million tokens.
For more complex reasoning tasks, developers turn to the DeepSeek V4 Pro model. This premium tier charges 0.025 RMB per million tokens for cache hits on the input side. A cache miss on the input side pushes the cost to 3 RMB per million tokens. Output tokens on the Pro model sit at 6 RMB per million tokens.
Because these rates are currently among the lowest in the artificial intelligence sector, even a moderate price hike will change developer budgets. Teams running high volume operations will feel the financial impact almost immediately. Users are advised to monitor their console alerts for the official rollout of the new pricing schedules.
