Vercel adds Google’s Gemini 4 Neon to AI Gateway

Vercel has made Google’s Gemini 4 Neon available on AI Gateway, adding a model aimed at coding, agent workflows and multi-step tasks. The model supports multimodal inputs, tool calling and a 1M-token context window.
- Gemini 4 Neon is now available on Vercel AI Gateway.
- The model supports text, images, video and documents.
- It has a 1M-token context window and four reasoning levels.
- Supported features include tool calling, structured outputs, Google Search and Google Maps grounding.
- The model ID is google/gemini-4-neon.
A faster model for coding and agents
Gemini 4 Neon is positioned as an upgrade over the earlier, lighter Gemini 4 model. According to Vercel, it offers output quality close to Google’s larger Gemini 4 models while providing faster responses.
Vercel says the model is built for coding, agent workflows and multi-step tasks. This places it among the models available through AI Gateway for applications that need to generate code, handle sequential operations or support agent-driven processes.
Multimodal inputs and a 1M-token context window
Gemini 4 Neon accepts text, images, video and documents. It has a 1M-token context window, allowing applications to work with large amounts of input within a single context.
The model also supports tool calling and structured outputs. For grounding, it can use Google Search and Google Maps. Developers can select among four reasoning levels: low, medium, high and xhigh.
How developers can access it
Developers can call the model using the ID google/gemini-4-neon. Vercel’s example uses the AI SDK and the streamText function to stream the model’s response while asking it to write a TypeScript function that debounces async calls.
Vercel says Gemini 4 Neon can also be used in Codex, Cursor or another supported agent. Developers are directed to the coding agents guide for setup information, and can try the model in the model playground or browse the available language models.
What the availability means for users
Adding Gemini 4 Neon to AI Gateway gives developers a single access point for a Google model designed for coding and agent use cases, alongside support for multimodal data, external grounding and structured responses. Its four reasoning levels also provide a way to choose how the model handles a task.
Vercel’s announcement does not provide pricing, performance figures or details about availability outside AI Gateway. The company’s claims about faster responses and output quality are comparisons with other Gemini 4 models described in the announcement.
Vercel: Gemini 4 Neon now available on AI Gateway


