Processing or generating 10,000 words (approximately 13,330 tokens) with DeepSeek V4 Flash Vision Exp ranges from $0.000187 (cached input) to $0.0176 (full generation).
Quick answer
For 10,000 words, DeepSeek V4 Flash Vision Exp processes approximately 13,330 tokens. That costs $0.005865 as uncached input, $0.000187 with the modeled cache discount, or $0.0176 when generated as output.
Method & trust
The estimate converts the selected English word count with the model tokenizer family, then applies the published input/output price per million tokens. Real prompts vary with code, formatting, language, and system instructions.
You send 10,000 words as prompt context, documentation, or background knowledge.
DeepSeek V4 Flash Vision Exp drafts a complete 10,000 words article, chapter, or code module.
8,000 words input prompt + 2,000 words output reply.
| Model | Provider | Input Cost (10,000 words) | Cached Input Cost | Output Cost (10,000 words) | Context Limit |
|---|---|---|---|---|---|
| DeepSeek V4 Flash Vision Exp (Current) | deepseek | $0.005865 | $0.000187 | $0.0176 | 1,050,000 |
| GPT-5.6 Luna | openai | $0.00256 | $0.000256 | $0.0154 | 1,050,000 |
| GPT-5.4 mini | openai | $0.0096 | $0.00096 | $0.0576 | 256,000 |
| GPT-5.4 nano | openai | $0.00256 | $0.000256 | $0.016 | 128,000 |
| GPT-4o mini | openai | $0.00192 | $0.00096 | $0.00768 | 128,000 |
| Claude Haiku 4.5 | anthropic | $0.0132 | $0.00132 | $0.066 | 1,000,000 |
For DeepSeek V4 Flash Vision Exp (deepseek_v4 tokenizer), 10,000 words is approximately 13,330 tokens (an average ratio of 1.33 tokens per word in English). Code, technical vocabulary, and non-English scripts will have higher token densities.
Sending 10,000 words as input costs $0.005865 (or $0.000187 with prompt caching). Generating 10,000 words as output costs $0.0176. Output tokens are more expensive because autoregressive token generation requires significantly more computation per token.
Yes! If your input text is part of a repeated context, prompt caching saves 50% to 90% (costing $0.000187). For non-urgent asynchronous batch processing, 24-hour batch queues provide a flat 50% discount across all tokens (costing $0.002933 for input, $0.008798 for output).
At an average human reading speed of 250 words per minute, 10,000 words takes approximately 40 minutes to read. In contrast, DeepSeek V4 Flash Vision Exp can process or generate this text in seconds.