DeepSeek V4 Pro
DeepSeekOfficial DeepSeek V4 API pricing for 2026. DeepSeek V4 Pro peak pricing: $1.32/M cache-miss input, $0.044/M cached input, $3.96/M output. Off-peak rates are 50% lower. 1M context, 384K max output.
Input Price
$1.32
cache miss / 1M tokens · Peak
Cached Input
$0.044
per 1M tokens · Peak
Output Price
$3.96
per 1M tokens · Peak
Context Window
1.0M
tokens
Off-peak: $0.66/M cache-miss input, $0.022/M cached input, $1.98/M output. All hours except 01:00-04:00 and 06:00-10:00 UTC.
DeepSeek V4 API Pricing 2026
New pricing took effect on August 16 at 16:00 UTC. The table shows off-peak to peak ranges per 1M tokens; peak hours are 01:00-04:00 and 06:00-10:00 UTC.
| Model | Cached input / 1M | Cache-miss input / 1M | Output / 1M | Best use |
|---|---|---|---|---|
| DeepSeek V4 Flash | $0.007-$0.014 | $0.22-$0.44 | $0.66-$1.32 | High-volume agent traffic |
| DeepSeek V4 Pro | $0.022-$0.044 | $0.66-$1.32 | $1.98-$3.96 | Harder coding and reasoning |
DeepSeek V4 API pricing: use cache-miss input for new prompt tokens and cached input for repeated prefixes.
DeepSeek V4 price: both models support OpenAI-compatible and Anthropic-compatible endpoints.
DeepSeek V4 cost: agent workloads can be far cheaper when repository context and system prompts hit cache.
Specifications
| Provider | DeepSeek |
| Model ID | deepseek-v4-pro |
| Input Price | $1.32 / 1M cache-miss tokens |
| Cached Input Price | $0.044 / 1M tokens |
| Output Price | $3.96 / 1M tokens · Peak |
| Off-peak | $0.66 / 1M cache-miss input, $0.022 / 1M cached input, $1.98 / 1M output · All hours except 01:00-04:00 and 06:00-10:00 UTC. |
| Context Window | 1.0M tokens |
| Max Output | 384K tokens |
| Capabilities | textfunction_callingstructured_output |
| Release Date | 2026-08-13 |
| Pricing Source | Official DeepSeek pricing |
| Price Verified | 2026-08-17 · New time-of-day pricing took effect at 2026-08-16 16:00 UTC. Canonical prices use the conservative peak rate; off-peak rates are recorded separately at 50% of peak. |
| Notes | Serves DeepSeek-V4-Pro-0813. New time-of-day pricing took effect on 2026-08-16 at 16:00 UTC. Peak hours are 01:00-04:00 and 06:00-10:00 UTC; off-peak rates are 50% lower. |
Monthly Cost Estimates
Estimated monthly costs based on different daily usage levels (assuming 50% input / 50% output split). Input estimates use cache-miss pricing, so cache-heavy workloads can be lower.
| Daily Tokens | Monthly Cost | Annual Cost |
|---|---|---|
| 10K | $0.792 | $9.50 |
| 50K | $3.96 | $47.52 |
| 100K | $7.92 | $95.04 |
| 500K | $39.60 | $475.20 |
| 1.0M | $79.20 | $950.40 |
About DeepSeek V4 Pro
DeepSeek V4 Pro is a large language model by DeepSeek. It features a 1.0M token context window with up to 384K tokens of output per request. The model supports 3 capabilities: text, function_calling, structured_output.
At $1.32 per million cache-miss input tokens and $3.96 per million output tokens, DeepSeek V4 Pro is positioned as a mid-range option in the DeepSeek lineup. Repeated prefix input can be charged at $0.044 per million cached tokens. Use our Token Counter to estimate how many tokens your prompts use, and our Pricing Calculator to compare costs across all models.
DeepSeek V4 Pro Key Details
- Pricing: $1.32/M cache-miss input tokens, $0.044/M cached input tokens, $3.96/M output tokens
- Context window: 1.0M tokens — one of the largest available
- Max output: 384K tokens per response
- Capabilities: text, function_calling, structured_output
- Highlights: Serves DeepSeek-V4-Pro-0813. New time-of-day pricing took effect on 2026-08-16 at 16:00 UTC. Peak hours are 01:00-04:00 and 06:00-10:00 UTC; off-peak rates are 50% lower.
- Released: 2026-08-13
Other DeepSeek Models
Similar Price Range
Related Tools
FAQ
How much does DeepSeek V4 Pro cost?
DeepSeek V4 Pro costs $1.32 per million cache-miss input tokens and $3.96 per million output tokens. Cached input costs $0.044 per million tokens. For a typical workload of 100K input tokens/day and 50K output tokens/day, expect approximately $9.90/month before cache-hit savings.
What is DeepSeek V4 Pro's context window?
DeepSeek V4 Pro supports a context window of 1.0M tokens. This means your combined input prompt and output response can be up to 1.0M tokens. The maximum output per response is 384K tokens.
Is DeepSeek V4 Pro good for my use case?
DeepSeek V4 Pro supports text, function_calling, structured_output. As a mid-range model, it balances capability and cost for most production use cases. Use our Pricing Calculator to compare with alternatives.
Is this the official DeepSeek V4 API pricing?
The table above is based on DeepSeek's official model pricing page, last checked against the current public docs. DeepSeek says product prices may change, so verify the official pricing page before committing large production spend.
Should I use DeepSeek V4 Flash or DeepSeek V4 Pro?
Use DeepSeek V4 Flash for high-volume chat, extraction, coding-agent subtasks, and cache-heavy repository work. Use DeepSeek V4 Pro for harder coding, reasoning-heavy agent tasks, and stronger long-horizon evaluations.
Does DeepSeek V4 have a free tier?
DeepSeek's public API docs describe billing from topped-up balance or granted balance, but they do not publish a permanent free tier table. Check your platform balance and the official docs for current granted credits before assuming free production usage.