OpenAI and Anthropic slash prices as Chinese open-weight models gain ground

The OpenAI price war is no longer a rumor. Both OpenAI and Anthropic have announced significant price cuts on their latest models, driven by competitive pressure from Chinese open-weight AI models that are gaining traction with cost-conscious US customers.

The price cuts in detail

OpenAI recently said it was slashing prices for GPT-5.6 Luna, its "fastest and most affordable model", by 80%. Anthropic responded by launching Claude Opus 5, touting "frontier intelligence at half the price" of Fable 5, its most capable model. These moves come as leading US AI labs release cheaper models to retain customers who are switching to cut-price alternatives from Chinese rivals.

Chinese open-weight models are the primary catalyst. Moonshot AI's Kimi K3 rivals Anthropic and OpenAI on price and has open weights, making it attractive for self-hosting. Demand was so high that Moonshot paused new Kimi K3 subscriptions after overwhelming compute capacity. Other Chinese labs like Z.ai are also upgrading their flagship models to compete directly with OpenAI and Anthropic.

What this means for AI builders

For teams building AI products, this is a direct reduction in API costs. An 80% cut on GPT-5.6 Luna makes it viable for high-volume use cases like customer support, content generation, and agent loops that were previously too expensive. Claude Opus 5 at half the price of Fable 5 brings frontier reasoning within reach for more applications.

But the OpenAI price war also introduces a strategic question: should you stick with US labs or evaluate Chinese open-weight models? The latter offer lower per-token costs and the ability to self-host, which can reduce dependency on a single provider. However, they come with trade-offs in data residency, compliance, and model safety. Mozilla's CTO switched to Kimi K3 for many daily tasks, signaling that even technical leaders are exploring alternatives.

Caveats to watch

The price cuts are vendor claims. OpenAI and Anthropic have not published independent benchmarks confirming that the cheaper models match the performance of their predecessors at the same reliability. Similarly, Chinese open-weight models may have different safety guardrails and are subject to geopolitical risks. The LA Times describes the situation as a "death zone" for anyone without frontier-pushing technology or market-breaking pricing.

For now, the OpenAI price war is a net positive for builders: lower costs, more options, and increased competition. The key is to evaluate models on your own workloads rather than relying solely on marketing claims.