DeepSeek
NewDeepSeek V4 Pro pricing
DeepSeek's April 2026 open-weight (MIT) flagship. A 1.6T-parameter mixture-of-experts model with a 1M-token context, priced far below Western frontier models.
Input
$1.74/ 1M tok
Output
$3.48/ 1M tok
- Context window
- 1M
- Max output
- 384K
- Cached input
- -
- Verified
- 2026-07-01
Standard vs promotional pricing
| Tier | Input / 1M | Output / 1M | Cached in / 1M |
|---|---|---|---|
Standard List rate | $1.74 | $3.48 | - |
Promo (75% off) Periodic promotional window | $0.43 | $0.87 | - |
DeepSeek periodically runs a 75% promotional discount that drops V4 Pro to $0.435 input / $0.87 output per 1M tokens. Confirm the current rate on the DeepSeek pricing page before you rely on it.
DeepSeek V4 Pro is the flagship of the DeepSeek V4 family, released on 24 April 2026 as an open-weight (MIT) model. It is a 1.6T-parameter mixture-of-experts architecture with roughly 49B parameters active per token, and it expanded the context window 8x over V3 – from 128K to a full 1M tokens – using a hybrid attention scheme (Compressed Sparse Attention plus Heavily Compressed Attention) built for long-context efficiency.
Pricing is the headline story. At $1.74 input / $3.48 output per 1M tokens it sits far below Western frontier models, and during DeepSeek's periodic 75%-off promotional windows it drops to $0.435 / $0.87 – roughly an order of magnitude cheaper than a flagship like Claude Opus or GPT-5.6 Sol. Max output is a generous 384K tokens.
Note that DeepSeek is not yet wired into the Calcis estimator or billing; these rates are informational and dated. The legacy deepseek-chat and deepseek-reasoner aliases are being retired on 24 July 2026 – new integrations should target deepseek-v4-pro (or V4 Flash) directly.
Estimate LLM costs before you send
Paste your prompt into the Calcis estimator to see token counts and per-request cost across every tracked model, then compare DeepSeek V4 Pro against them side by side.
Frequently asked
- How much does DeepSeek V4 Pro cost per 1M tokens?
- $1.74 for input and $3.48 for output per 1M tokens at the standard rate. During DeepSeek's periodic 75%-off promo, it drops to $0.435 input / $0.87 output.
- What is the DeepSeek V4 Pro context window?
- 1M tokens, with up to 384K output tokens - an 8x increase over the 128K context of DeepSeek V3.
- How does V4 Pro compare to V4 Flash on price?
- V4 Flash is dramatically cheaper at $0.14 input / $0.28 output per 1M tokens, aimed at high-volume and latency-sensitive work. V4 Pro is the flagship for the hardest tasks.
- Is DeepSeek V4 Pro open weight?
- Yes. The DeepSeek V4 family is released under the MIT licence as open-weight mixture-of-experts models, so you can self-host in addition to using the hosted API.
- Which model IDs should I use?
- Use deepseek-v4-pro directly. The older deepseek-chat and deepseek-reasoner aliases are scheduled for retirement on 24 July 2026.
Pricing verified 2026-07-01 from the provider's rate card. These figures are informational and not yet wired into the Calcis estimator or billing.