AI Coding Models Race to the Bottom on Price as Claude, GPT and Kimi Launch

Three major AI models — Anthropic's Claude Sonnet 5, OpenAI's GPT-5.6 family, and Moonshot AI's open-source Kimi K3 — launched within roughly one month, with pricing emerging as the defining battleground. Top-tier model costs have fallen to nearly a quarter of what they were not long ago, with competitive models now priced as low as $1 per million input tokens. The arrival of Kimi K3, a 2.8-trillion-parameter open-weight model released July 16, adds further downward pressure by threatening to push prices toward zero for those willing to self-host. Analysts note that cost-conscious enterprise buyers are now actively scrutinizing AI spending, making cheaper models increasingly attractive and giving vendors strong incentive to cut prices to capture developer loyalty. The AI coding market, estimated at around $4 billion annually, faces margin compression as the price war intensifies, raising questions about long-term profitability for even the largest players.