DeepSeek API Pricing
Open-weight AI models with the lowest prices in the industry — V4 Pro, V4 Flash, and V3
| Model | Input (per 1M tokens) | Output (per 1M tokens) | Context Window | Tier |
|---|---|---|---|---|
| DeepSeek V3.2 Deprecated | $0.23 | $0.34 | 128K | Budget |
| DeepSeek V4 Pro | $0.66 | $1.98 | 1M | Budget |
| DeepSeek V4 Flash | $0.22 | $0.66 | 1M | Budget |
| DeepSeek V3 Deprecated | $0.27 | $1.10 | 128K | Budget |
⏰ Peak vs Off-Peak Pricing
Off-peak: All hours outside peak times + all day Saturday & Sunday (Beijing time). Prices shown above are off-peak.
Peak hours: 01:00–04:00 and 06:00–10:00 UTC, Monday–Friday. Peak prices are exactly 2× off-peak.
V4 Flash peak: $0.44/$1.32 · V4 Pro peak: $1.32/$3.96
Why DeepSeek?
Chinese-built open-weight models that consistently offer the lowest prices per token in the industry.
Lowest Prices Available
DeepSeek V4 Flash at $0.22/$0.66 per 1M tokens (off-peak) is among the cheapest quality models available. V4 Pro at $0.66/$1.98 (off-peak) competes with GPT-5.4 at a fraction of the cost. Both have 1M context windows.
Open-Weight Models
All DeepSeek models are open-weight, meaning you can self-host them on your own infrastructure. No vendor lock-in — switch between API providers or run your own inference.
Strong Performance
DeepSeek V4 Pro rivals GPT-5.4 and Claude Sonnet 5 on major benchmarks while costing significantly less. V4 Flash offers near-instant responses for high-throughput workloads.
Which DeepSeek Model Should You Use?
Three models for different performance and cost requirements.
DeepSeek V4 Pro
- 1M context window
- Strong reasoning and coding
- Complex multi-step tasks
- Document analysis
- Rival quality to GPT-5.4
DeepSeek V4 Flash
- 1M context window
- Ultra-fast response times
- Classification and extraction
- High-volume workloads
- Cheapest quality model
DeepSeek V3 Deprecated
⚠️ Superseded by DeepSeek V4 Flash ($0.22/$0.66, 1M context). Switch to save 48% and get 8x larger context.
- 128K context window
- Proven and stable
- Good general-purpose model
- Wide ecosystem support
- Budget-friendly
DeepSeek Cost Calculator
Estimate your monthly spend on DeepSeek models.
Monthly Cost Estimate
Based on published API pricing. Actual costs may vary.
Need the full calculator with all providers? Compare DeepSeek against OpenAI, Anthropic, Google, and more.
Try the Full APIpulse CalculatorDeepSeek vs Competitors
How DeepSeek pricing stacks up against other major LLM providers.
| Provider | Model | Input (per 1M tokens) | Output (per 1M tokens) | Context Window |
|---|---|---|---|---|
| DeepSeek | V4 Pro | $0.66 | $1.98 | 1M |
| OpenAI | GPT-5.4 | $2.50 | $15.00 | 400K |
| Anthropic | Claude Sonnet 5 | $2.00 | $10.00 | 1M |
| Gemini 2.5 Pro | $1.25 | $10.00 | 1M | |
| DeepSeek | V4 Flash | $0.22 | $0.66 | 1M |
| OpenAI | GPT-5.4 nano | $0.20 | $1.25 | 400K |
| Gemini 2.5 Flash-Lite | $0.10 | $0.40 | 1M | |
| OpenAI | GPT-oss 20B | $0.08 | $0.35 | 128K |
See how DeepSeek models stack up against OpenAI, Anthropic, and more
Related Reading
Deep dives into budget AI pricing and cost optimization.
GPT-5.4-nano vs DeepSeek V4 Flash: Cheapest AI APIs Compared
DeepSeek V4 Flash ($0.22/1M off-peak) vs GPT-5.4-nano ($0.20/1M): which budget AI API should you use? Real benchmarks and trade-offs.
Compare them →Gemini 3.5 Flash vs DeepSeek V4 Flash: 95% Cheaper Budget AI
Gemini 3.5 Flash ($1.50/$9) vs DeepSeek V4 Flash ($0.22/$0.66). Same 1M context, 95% cheaper. Which budget model wins?
Compare them →The Cheapest LLM APIs in 2026: A Complete Ranking
Ranking every major LLM API by price per token. DeepSeek and GPT-oss models dominate the budget tier.
Read the full ranking →AI API Context Windows in 2026
DeepSeek V4 Pro and Flash both offer 1M context at budget prices ($0.66 and $0.22 input off-peak). Full comparison of every model's context window.
Read the full guide →Calculate Your DeepSeek Costs
See exactly what DeepSeek will cost for your specific workload. Compare against other providers in seconds.
Go to Calculator →Get notified when DeepSeek prices change
No spam. Only pricing updates and new features. Unsubscribe anytime.