ToolsSeptember 9, 2026via Vercel Blog
DeepSeek V4.1 Flash now available on AI Gateway
Why it matters
DeepSeek's fast multimodal model is now accessible through a unified API layer (Vercel's AI Gateway), lowering friction for practitioners building with multiple models. Vision support + 1M context + agent-framework integration (Cursor, Claude Code) makes this a practical addition to the toolkit, not just a model release.
Key signals
- DeepSeek V4.1 Flash now available on Vercel AI Gateway
- Vision/multimodal support (text + images in same request)
- 1 million token context window
- 384,000 token response limit
- Supports reasoning, tool use, prompt caching
- Available in Cursor, Claude Code, and coding agents via latest Vercel CLI
- No platform markup on inference; bring-your-own-key (BYOK) supported
- Separate input/output processing architecture reduces compute per stage
- DeepSeek V4.1 Flash supports native image understanding (vision capability)
- 1 million token context window; responses up to 384,000 tokens
- Reasoning, tool use, and prompt caching supported
- New architecture: separate input/output processing components reduce active computation
- Available on Vercel AI Gateway with no platform fee on inference, provider pricing passed through
- Integration with Cursor, Claude Code, and other coding agents via latest Vercel CLI
- Zero Data Retention support and custom reporting features included
- Model name for API calls: deepseek/deepseek-v4.1-flash
The hook
DeepSeek V4.1 Flash lands on Vercel's AI Gateway with vision — another model option for agent builders.
is now available on AI Gateway with native image understanding.DeepSeek V4.1 Flash
V4.1 Flash has vision support and accepts text and images in the same request, so you can ask questions about screenshots, read charts, and extract information from visual content.
The model has a 1 million token cont…