AI Agent News/stories/moonshot-s-kimi-k3-1-pro-and-highspeed-now-available-on-vercel-ai-gateway.md

Infrastructure

Moonshot's Kimi K3.1 Pro and Highspeed now available on Vercel AI Gateway

Vercel has added Kimi K3.1 Pro from Moonshot AI to AI Gateway, along with Kimi K3.1 Pro Highspeed, a faster version for interactive use. Moonshot says the model improves on Kimi K3 in coding, 3D modeling, financial analysis and agentic tasks while keeping its 1M-token context window and image input.

Source: vercel.com/changelog/kimi-k3-1-pro-and-kimi-k3-1-pro-highspeed-are-available-now-in-ai-gateway

Vercel has added Kimi K3.1 Pro from Moonshot AI to AI Gateway, together with Kimi K3.1 Pro Highspeed, a faster variant aimed at interactive use. Moonshot describes K3.1 Pro as the first model in its K3.1 series, built on Kimi K3 with gains in coding, 3D modeling, financial analysis and agentic tasks. It keeps the 1M-token context window and image input.

Vercel says the model handles everything from small fixes in an existing codebase to full projects, including long agent runs with many steps. It produces front-end pages and interactive scenes with stronger typography and animation than earlier Kimi models, and makes fewer factual errors on difficult prompts. In finance it can carry tasks such as a linked financial model with adjustable assumptions, or research grounded in source documents.

Developers can call the model through the AI SDK, the Chat Completions, Responses and Anthropic Messages APIs, or a coding agent connected to AI Gateway, using the identifier moonshotai/kimi-k3.1-pro. The Highspeed version uses moonshotai/kimi-k3.1-pro-highspeed. Moonshot plans to release the model weights by November 4, and Vercel says support for more providers will follow.

Why it matters if you build agents

Agent builders can now route coding and finance workloads to Kimi K3.1 Pro through AI Gateway with one model identifier, and use the Highspeed variant where responsiveness matters more.

AI Gateway says it passes provider pricing through with no markup and no platform fee on inference, including BYOK requests, so costs follow the provider's rates.

The weights are not yet public, so self-hosting the model is not possible until Moonshot's planned November 4 release.

Get stories like this every weekday

One short email with the day's agent tooling news.

Subscribe