DeepSeekopen weightsDeepSeek V3.1 pricing
$0.25 per 1M input tokens and $0.95 per 1M output tokens. In real work that is 0.60¢ per chat answer and 0.51¢ to summarize a 20-page PDF.
What DeepSeek V3.1 costs per task
Token counts estimated from a measured model of the same lab (DeepSeek V4.1 Flash), rescaled by tokenizer. Reasoning tokens are billed as output.
| Task | Cost | Tokens used | Price rank | Basis |
|---|---|---|---|---|
| Answer a chat question | 0.60¢$6.02 / 1,000 | 65 in · 859 out · 5,461 reasoning | 238 of 336 | estimated |
| Reply to a customer email | 0.19¢$1.95 / 1,000 | 279 in · 225 out · 1,749 reasoning | 230 of 336 | estimated |
| Write a 1,500-word blog post | 0.29¢$2.91 / 1,000 | 81 in · 2,230 out · 809 reasoning | 90 of 336 | estimated |
| Write 10 product descriptions | 0.27¢$2.65 / 1,000 | 307 in · 870 out · 1,842 reasoning | 97 of 336 | estimated |
| Translate 1,000 words | 1.1¢$11.46 / 1,000 | 1,354 in · 1,417 out · 10,291 reasoning | 190 of 336 | estimated |
| Proofread one page (500 words) | 0.27¢$2.67 / 1,000 | 746 in · 660 out · 1,954 reasoning | 176 of 336 | estimated |
| Summarize a 20-page PDF | 0.51¢$5.13 / 1,000 | 12,970 in · 427 out · 1,565 reasoning | 131 of 336 | estimated |
| Extract data from an invoice | 0.28¢$2.84 / 1,000 | 395 in · 449 out · 2,436 reasoning | 190 of 336 | estimated |
| Analyze a 200-row spreadsheet | ≥1.6¢$16.40 / 1,000 | 4,801 in · 0 out · 16,000 reasoning | 144 of 336 | estimated |
| Answer from 5 retrieved documents (RAG) | 0.45¢$4.49 / 1,000 | 3,696 in · 648 out · 3,102 reasoning | 161 of 336 | estimated |
| Classify 100 support tickets | 0.50¢$5.00 / 1,000 | 1,296 in · 461 out · 4,462 reasoning | 142 of 336 | estimated |
| Review 500 lines of code | ≥1.6¢$16.40 / 1,000 | 4,801 in · 0 out · 16,000 reasoning | 191 of 336 | estimated |
| Write a SQL query | 0.21¢$2.06 / 1,000 | 145 in · 406 out · 1,723 reasoning | 161 of 336 | estimated |
| Run a 50-step coding agent | 40¢$398 / 1,000 | 2,235,262 in · 14,985 out · 87,450 reasoning | 150 of 336 | modeled |
Price at every provider
List prices per 1M tokens (input / output) at each host serving DeepSeek V3.1 through OpenRouter today. Hosts can differ in quantization, context and speed.
| Provider | Input / output | Cached input | Context | Uptime |
|---|---|---|---|---|
 DeepInfrafp4 | $0.25 / $0.95 | $0.13 | 163K | 100% |
 SiliconFlowfp8 | $0.27 / $1.00 | - | 163K | 100% |
 SambaNovafp8 | $0.65 / $1.50 | - | 131K | 100% |
| C CoreWeavefp8 | $0.55 / $1.65 | $0.55 | 161K | 100% |
| M Maraunknown | $0.60 / $1.70 | - | 131K | 100% |
Price history
First seen in our price history on Aug 25, 2025 at $0.20 input / $0.80 output. Since then the output price moved from $0.80 to $0.95 per 1M tokens (13 changes).
FAQ
How much does DeepSeek V3.1 cost?
DeepSeek V3.1 costs $0.25 per 1 million input tokens and $0.95 per 1 million output tokens, with cached input at $0.13 (list price on Oct 18, 2026). In practice one chat answer costs about 0.60¢ and summarizing a 20-page PDF about 0.51¢.
Where is DeepSeek V3.1 cheapest?
Among 5 providers we track, the lowest list price today is at DeepInfra: $0.25 input and $0.95 output per 1M tokens.
Is the DeepSeek V3.1 cost per task measured or estimated?
Estimated: DeepSeek V3.1 was not run directly. Tokens come from DeepSeek V4.1 Flash and are rescaled by the tokenizer of DeepSeek. Real usage can differ.