What To Know
- The trend could have significant implications not only for Anthropic but for an AI industry that has invested billions of dollars in the assumption that customers will continually migrate towards bigger and more sophisticated models.
- Spending on Anthropic’s Fable 5 has reportedly plateaued at about 11 per cent of overall corporate expenditure on the company’s AI tools, based on payment data covering approximately 70,000 businesses.
Anthropic’s most powerful artificial intelligence model is facing an unexpected challenge: businesses appear increasingly reluctant to pay premium prices for frontier-level performance when cheaper AI systems can handle much of their everyday work. The trend could have significant implications not only for Anthropic but for an AI industry that has invested billions of dollars in the assumption that customers will continually migrate towards bigger and more sophisticated models.
Image Credit: Thailand AI News
Spending on Anthropic’s Fable 5 has reportedly plateaued at about 11 per cent of overall corporate expenditure on the company’s AI tools, based on payment data covering approximately 70,000 businesses. More than two months after the model’s release, adoption has therefore been slower than might have been expected from a flagship launch. As this AI News report examines, price rather than raw intelligence is increasingly becoming a decisive factor. Older Anthropic models and lower-cost alternatives can already perform many routine coding, analytical, customer-service and automation workloads without requiring companies to pay frontier-model prices.
Businesses Question the Frontier AI Premium
The development challenges one of the fundamental assumptions behind the economics of frontier AI. Developers including Anthropic have committed enormous amounts of capital to training increasingly capable systems, expecting improved intelligence to generate corresponding commercial demand.
Yet corporate customers are becoming more selective. One company director, whose firm has invested heavily in Anthropic, argued that most users simply do not need frontier capabilities for every task. Cutting-edge models remain important for demonstrating technological leadership and tackling exceptionally difficult problems, but they may increasingly become showcase products rather than the default choice for routine enterprise workloads.
Fable 5 also encountered political complications around its launch, including temporary US government restrictions linked to national-security concerns. Those uncertainties subsequently eased, leaving cost and practical performance as more important considerations for corporate buyers.
Anthropic Keeps Growing Despite Fable Slowdown
The weaker-than-expected Fable adoption comes at a sensitive moment for Anthropic as investors consider the company’s future growth and potential public-market ambitions. Its annualized revenue reportedly reached $65 billion in July, compared with $47 billion in May, although that remained below some bullish investor projections.
Nevertheless, the broader business continues expanding rapidly. Revenue has reportedly increased almost sevenfold since the beginning of the year, while Anthropic recorded an adjusted operating profit during the second quarter. The company has also told investors that around 6,000 customers spend at least $100,000 annually on its services.
Those figures suggest Anthropic’s central problem is not necessarily demand for AI. Instead, the emerging battle concerns which models customers choose and how much they are prepared to pay.
OpenAI and Chinese Models Intensify Price War
Competition is making that decision increasingly complicated. OpenAI has gained momentum following the introduction of GPT-5.6, while Anthropic’s cheaper Opus 5 has reportedly overtaken Fable 5 in business spending since its late-July launch.
An even greater pricing challenge is emerging from China. Models including Moonshot AI’s Kimi K3, Zhipu AI’s GLM-5.2 and GLM-5.3, Alibaba’s Qwen3.8-Max-Preview and lower-cost DeepSeek systems are attempting to narrow the performance gap while dramatically reducing inference costs.
The economics can be striking. Fable 5 has been cited at approximately $10 per million input tokens and $50 per million output tokens. By comparison, Kimi K3 has been listed around $3 and $15 respectively, while GLM-5.2 has been offered at roughly $1.40 for input and $4.40 for output. Highly optimized Chinese models can be cheaper still.
For businesses processing billions of tokens, such differences can translate into substantial operating savings.
AI Economics Could Be Entering a New Phase
The bigger question is whether frontier intelligence can continue commanding frontier prices. Companies may reserve premium models for complex reasoning, advanced research and difficult agentic tasks while routing ordinary workloads through cheaper systems.
If that behavior becomes standard, AI laboratories could face pressure to rethink how they recover enormous training and infrastructure investments. Anthropic’s Fable 5 therefore represents more than one model experiencing sluggish adoption; it highlights an industry-wide tension between technological supremacy and commercial efficiency.
Ultimately, the winners may not simply be the companies producing the smartest models. They could be those offering the strongest combination of capability, reliability and cost, particularly as increasingly competitive Chinese systems expand the choices available to global businesses. That shift could make AI pricing almost as important as AI intelligence in determining the next generation of market leaders.
For more on the better yet cheaper Chinese AI Models, visit:
Moonshot AI’s Kimi K3
Zhipu AI’s GLM-5.2 and GLM-5.3
Alibaba’s Qwen3.8-Max-Preview
https://qwen.ai/blog?id=qwen3.8
