DeepSeek V4 Pro
- API model IDs
- deepseek-v4-prodeepseek-v4-pro-0813
- Context window
- 1,000,000 tokens
- Max output
- 384,000 tokens
- Released
- 2026-08-13
- Lifecycle
- active
- License
- open-weights
Lifecycle verified 2026-09-08 (0 days ago). Source
Data handling
- Trains on API data by default
- Yes
- Retention
- Not published
- Zero data retention
- no
- Residency
- cn
- HIPAA BAA
- Not available
- SOC 2 / FedRAMP
- SOC 2 not published, FedRAMP none
DeepSeek's privacy policy states personal data, including prompts, is directly collected, processed, and stored in the People's Republic of China, and may also be stored on servers outside a user's home country. Users have a stated right to opt out of having their data used for training or optimizing DeepSeek's technologies; the policy does not describe this as off by default for API traffic, so training is recorded as on by default here unless a customer opts out. No specific retention period in days is published, and no zero data retention offering, HIPAA BAA, SOC 2, or FedRAMP authorization is stated.
Verified 2026-09-08 (0 days ago). Source
Pricing snapshot
$1.32 input, $3.96 output, per million tokens. Cached input $0.044 per million tokens.
This is a snapshot verified on 2026-09-08, 0 days ago, not a live price. Confirm on the provider’s pricing page before you budget against it.
Where you can call it
- direct · deepseek-v4-pro First-party DeepSeek API.
- self-host Weights published on Hugging Face under the MIT license.
DeepSeek runs a standing off-peak discount on top of the standard rate recorded here: prices are halved on cache-hit input, cache-miss input, and output tokens outside peak hours of 01:00 to 04:00 and 06:00 to 10:00 UTC, Monday through Friday. The price fields in this dataset record the standard (peak) rate; the off-peak halving is not a separately documented pricing field.