Chinese AI Firms Now Slashing Plans Like Anthropic

Chinese AI model providers are now adopting the same aggressive subscription downgrade tactics that earned Anthropic widespread criticism in Western markets. A major China-based AI company — identified by users only as starting with 'G' — recently slashed its coding plan limits while calling the replacement an 'equivalent' tier.

The backlash has been swift and pointed, with one veteran developer summarizing the community sentiment: 'Anthropic doesn't treat users like humans, and now you don't either?'

What Changed: The Numbers Tell the Story

The provider updated its user agreement on April 22, pulled legacy plans on April 30, and many users only discovered the changes via a console pop-up days later on May 3. The downgrade hits heavy users hardest.

Here's what changed between the old and new coding plans:

• Old plan: 600 API calls per 5-hour window, no weekly cap
• New plan: 400 API calls per 5-hour window, 2,000 weekly cap
• Theoretical old weekly maximum: Up to 19,800 calls (for power users working extended hours)
• Realistic old weekly usage: ~7,200 calls (10 hours/day, 6 days/week)
• New weekly ceiling: 2,000 calls — a 72% reduction from even moderate usage patterns

The company offered 2 months of a supposedly 'equivalent' package as compensation. Users quickly pointed out the math doesn't add up.

Designed for a 9-to-5 Developer Who Doesn't Exist

One frustrated developer ran the numbers and discovered something telling. The new weekly cap of 2,000 calls breaks down to exactly 400 calls × 5 hours per day × 1 week — suggesting the product team designed the tier around a developer who works just 5 hours a day, 5 days a week.

Real-world coding sessions look nothing like this. Professional developers regularly log 10-12 hour days during sprints, especially when using AI-assisted coding tools that encourage rapid iteration. The mismatch between the plan design and actual developer workflows reveals either a fundamental misunderstanding of the user base or a deliberate cost-cutting measure disguised as a plan 'update.'

Making matters worse, the company had publicly promised in March 2025 to restore legacy users' rights to renew their original subscription tiers. That commitment lasted barely 6 weeks.

A Global Pattern: From Anthropic to China

This incident mirrors a broader trend across the AI industry. Anthropic has faced repeated criticism for throttling Claude Pro usage, with subscribers reporting degraded performance and reduced message limits without clear communication. Cursor, the AI coding editor, similarly drew backlash for adjusting its 'fast request' quotas.

Now Chinese AI providers are following the same playbook:

• Quiet policy changes buried in updated terms of service
• Short transition windows that give users little time to react
• 'Equivalent' replacements that are objectively inferior
• Broken promises about grandfathering legacy subscribers

The pattern suggests this isn't a regional phenomenon but an industry-wide strategy as AI companies grapple with the reality that inference costs remain high and unlimited or generous plans are unsustainable at current pricing.

Why This Matters for AI Developer Tools

The subscription bait-and-switch problem threatens to erode trust at a critical moment for AI-powered development tools. Developers who build workflows around specific rate limits face real productivity disruptions when those limits change without adequate notice.

For the broader market, this trend raises important questions. If AI coding assistants can't maintain consistent service tiers, enterprise adoption — where predictability is paramount — becomes a harder sell. Companies evaluating these tools now have to factor in subscription instability risk alongside technical capabilities.

The developer community's response has been clear: transparent pricing and reliable service commitments matter as much as model quality. AI companies that fail to honor their promises risk losing users to competitors willing to play fair — regardless of which side of the Pacific they operate on.