Technology news and discussion on TalkFuse.
OpenAI and Anthropic in price war as Chinese AI rivals gain ground
OpenAI and Anthropic are cutting prices for some AI models as businesses seek cheaper alternatives from Chinese competitors such as Moonshot and DeepSeek. The shift matters because it challenges the US labs’ earlier focus on premium performance, while increasing pressure to show that heavy AI investment can produce returns ahead of planned public listings.
OpenAI cut GPT-5.6 Luna’s prices by 80%, to $0.20 per million input tokens and $1.20 per million output tokens, while Anthropic launched Claude Opus 5 at half the price of its Fable 5 model. Prices paid for leading US models have fallen by almost a quarter since mid-July, according to Silicon Data, and companies including DoorDash and Airbnb are testing Chinese models to reduce costs. Headline token prices are not a complete measure of value, however, as more capable models may use fewer tokens or attempts to complete a task.
In short
- US AI labs cut prices amid Chinese competition
- OpenAI reduced one model’s prices by 80%
- Cheaper models are reshaping corporate AI spending
Steelman: both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
The competition from Chinese rivals represents healthy market dynamics that benefits end users and businesses through dramatically lower costs. When prices fall by a quarter in a few months, it demonstrates that AI capabilities are becoming more efficient and commoditised, which is what should happen as technologies mature. This price competition will drive innovation by forcing companies to compete on efficiency and capability rather than premium pricing, whilst making advanced AI tools accessible to smaller businesses and organisations that previously couldn't afford them.
The case against
The aggressive price competition threatens the sustainability of the innovation cycle that produced these advanced models in the first place. Training and improving cutting-edge AI systems requires substantial capital investment that can only be justified if companies maintain sufficient margins and revenue; a race to the bottom on pricing could undermine the ability of US laboratories to fund the research necessary to maintain technological leadership. Chinese competitors may benefit from different funding structures or cost bases that don't reflect the true economics of developing frontier AI, making this competition potentially unsustainable and strategically concerning.
Explained simply
OpenAI and Anthropic are US companies that build AI models, which can generate text, answer questions and help with tasks such as writing software. Businesses often pay to use these models, with costs linked to the amount of text processed.
Chinese firms including Moonshot and DeepSeek are offering alternatives, giving businesses more choice as they weigh cost against how well a model performs. Lower prices put pressure on the US companies to show that their models are worth paying for, while they seek returns on large investments in AI. A model’s price per unit of text does not tell the whole story: a more capable one may need fewer attempts to complete a task.