DeepSeek's flagship combines thinking and standard modes. Optimised for coding and search agents, with the local Q4 build occupying about 404GB.
Which GPU does DeepSeek V3.1 671B-A37B need?
The entry configuration is 3x H200, which delivers around 22 tokens per second. For more parallel requests and headroom on speed, go with 6x H200.
How much does DeepSeek V3.1 671B-A37B cost?
An hour of DeepSeek V3.1 671B-A37B runs from $30.73 to $68.74 depending on the GPU configuration. Billing is hourly: you pay for server time, not for tokens.
How do I call DeepSeek V3.1 671B-A37B from my code?
Once deployed you get an OpenAI-compatible endpoint: point any OpenAI client at the base_url you receive and pass deepseek-v3.1:671b as the model name. Code written against the OpenAI API needs no changes.
Does the data stay private?
Yes. DeepSeek V3.1 671B-A37B runs on a dedicated GPU server inside a private loop: requests and responses never reach the model vendor and are not used for training.
What context window and license does DeepSeek V3.1 671B-A37B have?
DeepSeek V3.1 671B-A37B holds 32K tokens of context per request and ships under the MIT license, so you can use it commercially on your own hosting terms.