Vercel AI Updates: October 9, 2026

1. AI Gateway Supports OpenAI’s Ultrafast Service Tier

Vercel. AI Gateway now supports OpenAI’s Ultrafast service tier for openai/gpt-6-astra and openai/gpt-6.1-sol, requested via serviceTier: 'ultrafast' in the AI SDK’s OpenAI provider options or service_tier in Chat Completions and Responses API calls. Ultrafast requests are billed at 6x the standard per-token rate, and requests pinned to unsupported regions such as the EU fall back to the standard tier and are billed at that tier’s rate. OpenAI recommends the Responses API over a persistent WebSocket for tool-heavy workflows to cut per-turn overhead. Source

2. StepFun’s Step 5 Preview Added to AI Gateway

Vercel. StepFun’s flagship Step 5 Preview is available as stepfun/step-5-preview, targeting agentic coding, professional knowledge work, and financial analysis. The model accepts text and image input with a 1M-token context window. It can be used through the AI SDK, the OpenAI-compatible Chat Completions endpoint, or coding agents such as Claude Code after running npx vercel ai-gateway setup. Source

3. FLUX 3 Image Available for Generation and Editing

Vercel. Black Forest Labs’ FLUX 3 Image is available on AI Gateway as bfl/flux-3-image, supporting text-to-image generation and editing with up to ten reference images passed in prompt.images. It offers five resolution tiers from 768x768 up to 4K across square, portrait, and landscape aspect ratios. Usage requires AI SDK 7 or later and an AI Gateway API key, with no separate Black Forest Labs account. Source

4. Grok Imagine Video 1.5 Lite Brings Video With Native Audio

Vercel. xAI’s Grok Imagine Video 1.5 Lite is available as spacexai/grok-imagine-video-1.5-lite, generating 1 to 15 second clips with native audio from text prompts or images. It supports 480p, 720p, and 1080p output across seven aspect ratios. Developers can call it with the AI SDK’s experimental_generateVideo, the AI CLI, or the AI Gateway Playground, using a gateway key or BYOK. Source