Vercel AI Updates: July 14, 2026

1. Chat SDK Adds an X Adapter

Vercel. Vercel’s Chat SDK gained an adapter for X, formerly Twitter, letting bots built on the SDK reply to public mentions and hold direct message conversations through the X API v2. The adapter extends the same conversational agent framework beyond web chat into a social platform, so a single Chat SDK backend can drive interactions across surfaces. Source

2. AI SDK xAI Provider Adds URL Validation for Provider Responses

Vercel. Vercel released @ai-sdk/xai@4.0.12, a security-focused patch that adds URL validation to provider responses. The getFromApi function now accepts a validateUrl flag that routes URLs through fetchWithValidatedRedirects, rejecting private, loopback, and link-local targets and validating each redirect hop. New credentialedOrigin and trustedOrigin parameters give developers control over where API keys are transmitted, which matters for self-hosted deployments. Source

3. AI Gateway Production Index Reports Open-Weight Models at 29% of Volume

Vercel. Vercel published its July 2026 AI Gateway Production Index, reporting that open-weight models ran 29% of gateway tokens in June while accounting for just under 4% of spend. The report notes that price per token has flattened across the market, and frames open-weight adoption as increasingly driven by cost efficiency rather than capability tradeoffs. Source

4. Agent Runs Surface Subagent Activity on eve Projects

Vercel. Vercel added a Subagents tab to Agent Runs on eve projects, letting teams inspect subagent activity through a shared timeline that shows each subagent’s prompts, duration, and failures. The added visibility helps developers debug multi-agent workflows where a top-level run delegates to nested agents. Source

5. Deployment Policies Restrict Which Sources Can Create Deployments

Vercel. Vercel introduced Deployment Policies, which let teams restrict the mechanisms, organizations, and repositories allowed to create deployments at both the team and project level. The controls give platform administrators a way to lock down deployment creation to approved sources, tightening supply-chain governance for production projects. Source