DeepSeek V3 Pricing
BalancedDeepSeek · 164K tokens context
DeepSeek V3 from DeepSeek costs $0.320 per 1 million input tokens and $0.890 per 1 million output tokens as of September 2026 (live OpenRouter data). The model supports a 163,840-token context window (approximately 122,880 words) with a 8K-token maximum output. A typical 1,000-token request costs $0.0003 in input charges; a 10,000-token request costs $0.0032.
| Input price | $0.320 / 1M tokens |
|---|---|
| Output price | $0.890 / 1M tokens |
| Output / input ratio | 2.8× |
| Context window | 163,840 tokens (~122,880 words) |
| Maximum output | 8,000 tokens |
| Cost per 1K tokens (input) | $0.0003 |
| Tier | Balanced |
| Last verified |
DeepSeek V3 is the general-purpose DeepSeek model — surprisingly strong on coding and math at a fraction of GPT-4o's price. Open-weight, with growing third-party host support.
Input Price
$0.320
per 1 million tokens
Output Price
$0.890
per 1 million tokens
Context Window
164K tokens
max 8K output
Cost Examples
| Request Type | Tokens | Input Cost | Output Cost |
|---|---|---|---|
| 1,000 word article | 1,333 | $0.000427 | $0.000356 |
| 10-page document (2,500 words) | 3,333 | $0.00107 | $0.00089 |
| 1,000 lines of code | 5,000 | $0.0016 | $0.00133 |
| 100K token document | 100,000 | $0.032 | $0.0267 |
Output cost estimated at 30% of input token count. Use the calculator for exact figures.
Strengths
- ✓Excellent quality-per-dollar
- ✓Strong coding benchmarks
- ✓Open weights
Limitations
- –Smaller context window than US-led models
- –Less mature ecosystem outside China
Best Use Cases
Calculate DeepSeek V3 Costs
Use the TokenRate calculator to convert any budget, token count, or text into exact DeepSeek V3 costs — and compare across all models.
Open Calculator →DeepSeek V3 — FAQ
Related Models
Related Guides