AI Product Updates: New Features and Releases

One API gives creators access to creative tools and several leading models at once.

SpaceXAI released a major Grok Build update that makes Grok 4.7 available in the product, according to a post shared by Musk.

Third-party testing shows GPT-6 Sol and Luna cost about half of GPT-5.6 at the same Intelligence Index, with Sol's coding agent score up slightly and Luna's down, and hallucination rates notably lower.

QuiverAI's Arrow 2 cuts API token rates for large-batch vector workflows, with usage-based billing across generation and refinement.

OpenAI's developer account says improved prompt caching in GPT-6 keeps context reusable as reasoning effort and tool availability change, making responses faster and cheaper.

GPT-6 Sol and Luna roll out in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu users, plus the API, with Luna available to Free and Go users in the desktop app.

OpenAI says higher usage limits and lower cost give users more flexibility and room to iterate, without specifying tiers in the post.

Colab is now part of Google AI plans, giving subscribers priority access to faster accelerators, with Ultra users also getting longer-running background execution and Premium GPU access.

Museverse grades AI agents' revenue claims under a "no proof = $0" rule, sorting them into unproven, unconfirmed, and verified tiers so only checkable figures count on the leaderboard.

Firecrawl, which scrapes web pages into structured data for AI agents, has raised a $75M Series B and launched Alexandria, a knowledge library product.

Anthropic's Claude Opus 5.5 has entered Agent Arena, where models are evaluated on real-world, long-horizon agentic tasks and ranked by user votes.

Agent Arena has opened test access to Opus 5.5 in Battle Mode and Agent Mode, with user votes affecting the leaderboard.