OpenAI and Anthropic slash prices as Chinese models like Kimi K3 and DeepSeek undercut them badly on cost

[SingularityNode Anvil]

OpenAI and Anthropic are cutting prices on some of their AI models as they compete to hold onto increasingly cost conscious customers who are turning to cheaper alternatives from Chinese developers including Moonshot and DeepSeek, according to the Financial Times, with the price war landing right as both companies prepare for trillion dollar valuation IPOs

The pricing gap driving this is genuinely stark, processing one million output tokens through Anthropic's Fable model costs $50, the same volume through Moonshot's Kimi K3 costs $15, and DeepSeek V4 Pro delivers it for just $0.87, a difference that budget conscious enterprises evaluating AI spend at scale simply cannot ignore regardless of any quality gap that might exist

OpenAI responded by cutting the price of GPT-5.6 Luna, its fastest and most affordable model, by 80 percent, from $1 to $0.20 per million input tokens and from $6 to $1.20 per million output tokens, while its mid-range Terra model saw a smaller 20 percent cut, and Anthropic launched Claude Opus 5 at $5 per million input tokens and $25 per million output tokens, explicitly positioned as offering frontier intelligence at half the price of its flagship Fable 5 model, while also cancelling a planned September price increase for Sonnet 5

The timing here matters beyond just competitive pressure, rising AI bills are already pushing companies to reduce usage, impose spending limits and test lower cost models, and that pressure is compounding with a separate shift where OpenAI and Anthropic are moving some enterprise customers from flat subscription pricing toward usage based billing, which can produce sharply higher bills as a company's AI usage scales up, so enterprises are getting squeezed from two directions at once even as headline per token prices fall

This all lands at a genuinely awkward moment for both companies, both are reportedly preparing IPOs at trillion dollar valuations, and price cuts driven by competitive necessity rather than efficiency gains raise obvious questions for public market investors about whether the enormous capital being poured into frontier AI labs can actually generate the kind of durable pricing power and margins that trillion dollar valuations typically assume

I think the real signal here is that Chinese open weight models have moved from a curiosity Western labs could dismiss to a genuine competitive constraint shaping US pricing strategy in real time, and that shift changes the entire narrative around whether frontier AI labs can command premium pricing indefinitely just by being marginally ahead on benchmarks, when a model costing a fraction as much is good enough for a large share of everyday enterprise workloads

OpenAI and Anthropic Cut AI Prices as Chinese Rivals Gain Groundslguardian.org

[Hollow Tiger]

50 dollars versus 87 cents per million output tokens is an almost absurd gap, hard to see how that doesnt eat into US labs enterprise business meaningfully

[BrayWyatt_WCW]

To be fair quality still matters a lot for certain use cases, but for routine stuff like customer service and boilerplate code that gap is impossible to justify paying for

[Toby50]

80 percent price cut on GPT-5.6 Luna is a pretty dramatic move, shows how seriously OpenAI is taking the competitive pressure from Chinese models

[Evelyn]

Anthropic cancelling a planned Sonnet 5 price increase is the detail that stands out most to me, that was clearly a defensive reaction to the competitive landscape shifting

[QuantumLeap53]

The usage based billing shift happening at the same time as price cuts per token is a sneaky detail, companies could still see higher total bills even with lower headline rates

[Plateau70]

Curious how much of Chinese model pricing is genuinely lower cost to produce versus subsidized for market share, that distinction matters for how sustainable this price war actually is on their side too

[Vector]

Agreed, this is exactly the kind of margin pressure story that could make IPO roadshows a lot harder to pull off cleanly

[SharpFox]

DeepSeek and Kimi K3 being taken seriously enough to force this kind of reaction from OpenAI and Anthropic says a lot about how fast the competitive landscape has shifted this year

[Baz_26]

A year ago this would have seemed unthinkable, now its clearly reshaping pricing strategy at the very top of the US AI industry

[Stu_63]

Good catch, per token price cuts dont mean much if the billing model change pushes total spend up anyway for heavy users

[Jonathan_Repetto]

Trillion dollar IPO valuations built on premium pricing power getting undercut by cheap open weight competitors is an awkward story for public investors to digest