Loading Legacy
DeepSeek V3.2 costs $0.28 per million input tokens and $0.42 per million output tokens on the DeepSeek API (as of 2026-07-13), with a 128K-token context window.
DeepSeek's open-weight model — among the cheapest API prices of any frontier-adjacent model.
Prices as of 2026-07-13 · verify on DeepSeek's pricing page
…
Tokens (o200k)
exact for GPT-family
28
Words
— tokens/word
145
Characters
— chars/token
| Model | Input tokens | Input $/1M | Output $/1M | Context | This request |
|---|
GPT-family counts are exact — they use the same open o200k tokenizer as the models. Anthropic and Google keep their tokenizers private, so those counts are clearly-labeled estimates (Claude typically tokenizes ~10–20% above o200k; use Anthropic's free count_tokens API endpoint for exact numbers). Prices come from each provider's public pricing page — follow the links on the model pages to verify. Everything runs in your browser; your text is never uploaded.
| Request | Input tokens | Output tokens | Cost |
|---|---|---|---|
| Short prompt | 500 | 500 | $0.00035 |
| Long prompt / article | 5,000 | 1,000 | $0.00182 |
| Small codebase / big doc | 50,000 | 4,000 | $0.02 |
| Full context window | 128,000 | 8,000 | $0.04 |