Anthropic reportedly blocks OpenClaw (小龙虾); Chinese models seize opening as tensions mount

Ban and backlash

Anthropic reportedly moved to cut off third‑party access to its Claude model, a decision that has provoked an outcry in AI communities after OpenClaw (小龙虾) — an app that helped make AI mainstream this year — relied on Claude subscriptions to power many users' features. OpenClaw became a viral phenomenon, and many users paid premium prices to route Claude through the product. Now some of those high‑price bets are backfiring as access is restricted, and it has been reported that Anthropic’s stance is firm and unlikely to be reversed.

Domestic models pounce

Domestic competitors were quick to respond. MiniMax (迷你马克斯), one of China’s fast‑rising large models, issued a pointed statement arguing that locking AI subscriptions into first‑party products “strangles” third‑party innovation — and at the same time highlighted that its token system was designed for cross‑platform use. It has been reported that MiniMax’s 2.5/2.7 variants have ranked among the top callers on OpenRouter for the past two months, giving Chinese providers a ready springboard to capture users displaced by Claude’s block.

Geopolitical ripple and what’s next

This episode is more than a platform spat. Western providers tightening control over model access comes amid broader US‑China technology frictions and export‑control debates that shape who gets cutting‑edge AI tools and how they’re used. Adding another twist, it has been reported that Google may open‑source its Gemma 4 model and make it broadly accessible to developers in mainland China — a move that, if confirmed, could reshape choices for developers asking: buy into paid subscriptions now, or wait for new, more open alternatives? The scramble reveals how quickly AI market dynamics can shift when platform gates are slammed shut.