Anthropic drops plan to charge extra for Fable 5 as Chinese rivals close in

Anthropic dropped plans to charge extra for Claude Fable 5 on July 20, making the model free for Max and Team Premium subscribers within their usage limits, three days after Chinese lab Moonshot AI released Kimi K3, a 2.8 trillion parameter model that beats Fable 5 on some benchmarks at less than a third of the price. Pro and Team Standard subscribers get a one-time $100 usage credit before facing metered pricing of $10 per million input tokens and $50 per million output tokens, as Chinese rivals including Moonshot, Zhipu, DeepSeek, and Alibaba's Qwen intensify price competition.

Anthropic just backed off making its best customers pay extra for Claude Fable 5, three days after a Chinese lab released a model that beats it on some benchmarks for less than a third of the price. That's a fast reversal. Anthropic said on July 20 that Fable 5 is staying inside its paid Claude plans, and for Max and Team Premium subscribers, that means no additional per-token charge on top of what they already pay. Those subscribers can burn through up to half their weekly usage limit on Fable 5 with zero metered billing, according to reporting from Winbuzzer and TechTimes. Three months ago, Anthropic was preparing to push every paying user onto metered pricing for the model. Now only some of them will pay it. Pro and Team Standard subscribers are not getting the same deal. They get a one-time $100 usage credit. Once that runs out, Fable 5 costs $10 per million input tokens and $50 per million output tokens, per TechTimes. That is a retreat with an asterisk, not a giveaway. Anthropic had already flip-flopped once before landing here. It planned to cut off included Fable 5 access on July 12, then extended the deadline to July 19 after developers who had built workflows around free access pushed back, according to Forbes and Bleeping Computer. A day after that second deadline passed, the company made the arrangement permanent for its top tier. Anthropic has told users the reversal reflects demand it struggled to manage and strain on its computing capacity. The timing tells its own story. Three days before Anthropic's July 20 announcement, Moonshot AI, a Beijing startup backed by Alibaba and Tencent, released Kimi K3, a 2.8 trillion parameter open-weight model that landed third on the Artificial Analysis Intelligence Index, behind only Fable 5 and OpenAI's GPT 5.6 Sol, CNBC reported on July 17. Kimi K3 did not just rank well. On several front-end coding and agentic benchmarks it beat Claude Opus 4.8 and GPT 5.5 outright, and on some coding tests it edged past Fable 5 and GPT 5.6 Sol themselves. And it's cheap. Moonshot priced K3's output at $15 per million tokens. Fable 5 charges $50 for that same million tokens under metered billing. That's less than a third of the price for a model beating Anthropic's own flagship on some of the benchmarks developers actually care about. Kimi K3 is not the only Chinese model closing the gap. Zhipu's GLM 5.2 leads the Chinese field on independent coding and agent leaderboards, DeepSeek V4 undercuts everyone on price, and Alibaba's Qwen has become the most widely adopted open base model among developers building their own tools, according to rankings compiled by BenchLM and GEO Toolbox. No single one of them threatens Fable 5's overall lead. Together, they are squeezing what a developer is willing to pay for the best model when a much cheaper one gets close enough. Fortune described the market reaction to Kimi K3 as a new Also read: South Korea's Record Exports Show How Deep Its AI Chip Bet Really Goes https://startupfortune.com/south-koreas-record-exports-show-how-deep-its-ai-chip-bet-really-goes/ • A federal judge just approved Anthropic's $1.5 billion settlement with authors https://startupfortune.com/a-federal-judge-just-approved-anthropics-15-billion-settlement-with-authors/ • Oracle Sues Wisconsin Over A $7 Billion Data Center Collateral Bill https://startupfortune.com/oracle-sues-wisconsin-over-a-7-billion-data-center-collateral-bill/