DeepSeek V4 Flash Vision Exp
Provider: DeepSeek CN
Overview
DeepSeek V4 Flash Vision Exp is an experimental vision-enabled reasoning model by DeepSeek built for multimodal tasks across a 1M token context.
Strengths
- 1048576 token context window with 943717 max output
- Native vision, function calling, and JSON mode support
- Explicit thinking capability included for complex agent logic
- Cached input drops sharply to $0.00686 per million
Best use cases
- High-volume document parsing with visual layout inputs
- Automated agent workflows requiring structured JSON outputs
- Long-context retrieval tasks across large codebases
- Multimodal analysis leveraging the 1M input window
Pricing position
Input sits at $0.2156 and output at $0.6468 per million tokens, closely matching MiniMax M2.7 and Qwen3 VL 30B.
Caveats
Third-party host pricing applies with no free tier available. Output cost is higher than the base V4 Flash 0731 variant.
Notes drafted by an LLM from the figures on this page (2026-11-10) and rewritten when the price changes.
Key facts
- Price basis
- Median third-party host price (no first-party API)
- Max output
- 944K tokens
- Listed on OpenRouter
- 2026-08-21
- Tracked since
- 2026-08-21
- Last updated
- 2026-11-10
Features
Price history
- 2026-08-21 Added
Monthly cost estimate
| Scale | Input (M) | Output (M) | Monthly |
|---|---|---|---|
| 0.1M (hobby) | 0.1 | 0.05 | $0.05 |
| 1M (small app) | 1 | 0.5 | $0.54 |
| 10M (mid scale) | 10 | 5 | $5.39 |
| 100M (large) | 100 | 50 | $53.90 |
※ Cache and batch discounts not factored in. Detailed calculator
Official resources
Other models from DeepSeek
Similar-priced models
Source: OpenRouter API (deepseek/deepseek-v4-flash-vision-exp)