OpenAI slashes ChatGPT prices by 80% after Kimi K3 takes the AI world by storm

OpenAI has just cut prices of its flagship GPT-5.6 family of AI models by as much as 80 per cent. This move comes at a time when US AI companies are facing pressure from cheaper models from China such as Moonshot's Kimi K3 and Zai's GLM 5.2.OpenAI has just made two of its models cheaper.OpenAI is making its flagship AI models cheaper. OpenAI CEO Sam Altman has announced that the AI startup is cutting prices of two models from the GPT-5.6 family – GPT-5.6 Luna and GPT-5.6 Terra. This move comes at a time when OpenAI is facing stiff competition from Chinese open-weight models like Kimi K3 that are cheaper to use.Sam Altman shared the update on X. “Major price cuts today: 80 per cent drop for GPT-5.6 Luna, now $0.20 (roughly Rs 20) per million input tokens and $1.20 (roughly Rs 115) per million output,” he wrote. “20 per cent drop for GPT-5.6 Terra, to $2 (roughly Rs 191) /$12 (roughly Rs 1,150).” Sam Altman announced the new prices on X. To give you some context, the price cuts for Luna and Terra make them cheaper than Kimi K3, which is the largest open-weight AI model out there. Kimi K3 is priced at $3 (roughly Rs 287) per million input tokens and $15 (roughly Rs 1,436) per million output tokens.Kimi K3 has shocked the global AI industry since its launch as it is able to match the performance of GPT-5.6 Sol and Anthropic’s Claude Fable 5 in certain tasks, but at a fraction of the cost. OpenAI launches fast mode for GPT-5.6 SolBut costs are not the only thing OpenAI is addressing. For those who want the best level of performance at a quicker pace, the AI startup has also launched a fast mode for GPT-5.6 Sol, its most advanced AI model.Sam Altman said, “GPT-5.6 Sol gets Fast mode in the API, up to 2.5x the speed for 2x the price, same intelligence.” That is, you get much faster responses from Sol but at double the price. “We want to offer the best price/intelligence tradeoff at every level,” Altman added. OpenAI is also making its models more efficient. The AI startup says that after deployment it used GPT-5.6 Sol to make itself more efficient to run, resulting in 20 per cent lower serving costs and more than 15 per cent better token-generation efficiency. The broader pricing shift comes as enterprises become more cost-sensitive and less willing to deploy expensive AI systems without a clear view of returns on investment.As AI bills have grown, in some cases into the billions of dollars, many businesses, such as Uber and Walmart, have started trying to control spending more closely. OpenAI and other US AI labs are also facing pressure from Chinese open-weight rivals, which can be downloaded, modified and run on a company’s own infrastructure. The weights of Kimi K3, for instance, are now available to download for free, allowing developers to run the model locally and tune it to their needs.OpenAI’s latest pricing also adds pressure on rivals including Anthropic and Google, all of which have recently highlighted cheaper or more cost-effective AI models. Meta and xAI have also launched more cost-efficient models recently. That is, we may see a price war brewing over AI models in the Silicon Valley as these companies try and compete for entreprise users.- EndsPublished By: Armaan AgarwalPublished On: Jul 31, 2026 08:36 IST