DeepSeek V4.1 Flash
DeepSeek V4.1 Flash with native visual understanding
DeepSeek V4.1 Flash uses deepseek-flash with native visual understanding for image-text Q&A, coding assistance and content analysis.
starsCapabilities
codeFunction callingdata_objectStructured output
paymentsContext and pricing
Context limit1,000,000
Max output384,000
Input price¥2/ 1M tokens
Output price¥8/ 1M tokens
Cached input price¥0.04/ 1M tokens
descriptionOverview
Overview
Released on September 10, 2026, DeepSeek V4.1 Flash uses official API model ID deepseek-flash and supports native visual understanding for image-text analysis, coding assistance and general Q&A.
Migration
The release announcement retires V4 Flash and Vision Exp and temporarily routes their compatibility IDs to V4.1 Flash. The announcement and pricing page conflict on Pro routing, so they do not establish the current backing version of the old Pro alias. Use deepseek-flash for this model. No independently unverified prices are recorded here.
lightbulbUse cases
- Image-text Q&A
- Screenshot and document analysis
- Coding assistance
- Content generation and summarization
thumb_upStrengths
- Native visual understanding
- Explicit official API model ID
- Migration target for legacy Flash compatibility routes
infoLimitations
- Visual understanding does not imply image generation
- Legacy compatibility routes may change
- Official documents conflict on Pro routing
- Check current official input limits and billing
Scan to join WeChat group