DeepSeek (深度求索) Pricing & Token Calculator
All 4 DeepSeek (深度求索) models on one page: context windows, tokenizer behaviour and list price per 1M input and output tokens. Each row links to a dedicated calculator already preselected to that model.
DeepSeek does not expose its tokenizer. TokenCalc estimates from character ratios calibrated on Chinese and English prompts.
DeepSeek (深度求索) model pricing table
| Model | Tokenizer | Context | Input / 1M | Output / 1M |
|---|---|---|---|---|
| DeepSeek V4 Pro | estimate | 128K | $0.44 | $0.87 |
| DeepSeek V4 Flash | estimate | 400K | $0.14 | $0.28 |
| DeepSeek V3.2 | estimate | 128K | $0.35 | $0.42 |
| DeepSeek R1 | estimate | 64K | $0.56 | $2.50 |
Cheapest DeepSeek (深度求索) models per 1M input tokens
ERNIE X1 at $0.006/1M, Doubao 1.6 Flash at $0.01/1M, text-embedding-3-small at $0.02/1M, Doubao 1.6 Lite at $0.02/1M, SenseNova 5.1 (商汤) at $0.04/1M. Output rates run several times higher than input rates on most vendors, so a token-heavy workflow with long generations is priced by the output column.
Largest DeepSeek (深度求索) context windows
Gemini 3.1 Pro (2M), Gemini 2.5 Pro (2M), GPT-5.5 (1M). Context headroom is usually the binding constraint before price is — check it first.
Count tokens for any DeepSeek (深度求索) model
Pick the model from the dropdown, or open any row above for a page preselected to it.
DeepSeek (深度求索) pricing FAQs
How many DeepSeek (深度求索) models can I count tokens for?
4 on TokenCalc, each with its own page and its own preselected calculator: DeepSeek V4 Pro, DeepSeek V4 Flash, DeepSeek V3.2, DeepSeek R1.
Are DeepSeek (深度求索) token counts exact?
DeepSeek does not expose its tokenizer. TokenCalc estimates from character ratios calibrated on Chinese and English prompts.
Which DeepSeek (深度求索) model is cheapest per 1M input tokens?
Among the models tracked here: ERNIE X1 at $0.006/1M, Doubao 1.6 Flash at $0.01/1M, text-embedding-3-small at $0.02/1M, Doubao 1.6 Lite at $0.02/1M, SenseNova 5.1 (商汤) at $0.04/1M. Prices verified 2 October 2026.