OpenAI and Anthropic have reduced prices for their flagship AI models by as much as 80% over the past four months, as competition from cheaper open-weight models intensifies and both companies face questions about potential public listings.
The rapid price cuts, which began in June, have lowered API usage costs by 75% or more in some cases. Competitors such as DeepSeek and Moonshot AI have increased pressure on the two companies by offering lower-cost models through platforms such as OpenRouter, where dozens of providers offer AI inference services.
Major Price Cuts
OpenAI began the latest round of reductions in late June when it launched GPT-5.6 Luna at $1 per million input tokens and $6 per million output tokens. By July 30, the company had reduced both prices by 80% to $0.20 and $1.20, respectively.
Anthropic followed in August by making its temporary launch discount for Claude Sonnet 5 permanent. In September, it further reduced Claude Opus 5.5 pricing by 40% to 60%.
Days later, OpenAI introduced GPT-6.1 Sol at half the input and output prices of its predecessor. The company also reduced cached input pricing to 5% of the regular input cost, matching Anthropic’s rate.
Max Woolf, a former senior data scientist at BuzzFeed who tracks AI model pricing on his blog, said the speed of the reductions surprised him. He noted that the cost of using leading AI models has fallen much faster than he expected a year ago.
The lower prices are particularly important for AI coding tools such as Codex and Claude Code. These tools repeatedly access cached conversation histories, with cache reads accounting for 96% to 99% of token spending in such workflows.
As a result, a 75% reduction in cache-read prices can reduce effective usage costs by approximately 70% to 75%.
Data from Vercel’s AI model gateway also shows that the average cost per token has fallen to 43.3% of its April 2026 level. However, total spending continues to increase because improvements in AI model capabilities are driving greater usage.
Cheaper Open-Weight Models Increase Competition
OpenAI and Anthropic are also facing stronger competition from open-weight AI models, particularly those developed by DeepSeek and Moonshot AI.
DeepSeek’s Flash V4.1 1017 model, released in April, offers input, output, and cached token pricing of $0.15, $0.60, and $0.003 per million tokens, respectively.
The model is now available through 29 different inference providers on OpenRouter, compared with just 13 in May. Providers such as Cloudflare and DeepInfra offer effective usage costs within 10% of DeepSeek’s own endpoint.
Meanwhile, Moonshot AI’s Kimi K3.1 Pro reportedly delivers performance comparable to Claude Opus 5.5 at half the input and output price. Moonshot AI plans to release the model’s weights on November 4.
Pricing comparisons also show that OpenAI’s 80% reduction for GPT-5.6 Luna significantly improved its cost competitiveness against DeepSeek V4 Pro. Without the reduction, DeepSeek’s model would have offered a better position on the reported quality-versus-cost comparison.
Despite their lower prices and growing availability on OpenRouter, open-weight models still account for a minority of total token usage.
Woolf said OpenAI and Anthropic must continue developing higher-quality models to compete with cheaper alternatives. He identified Claude Opus 5.5 as an example of how premium models can maintain an advantage through performance.
Price Cuts Raise Questions About Potential IPOs
The reductions also come as both companies reportedly consider entering the public market.
OpenAI is rumored to be targeting an initial public offering in 2027, while Anthropic could file for an IPO within the next few months.
Both companies have reported billions of dollars in losses related to research, development, and infrastructure expansion. However, those figures predate the recent increase in AI agent usage.
Cutting prices by as much as 80% could complicate their financial positioning ahead of potential IPOs, particularly if lower prices reduce profit margins.
However, some reports suggest the companies previously generated profit margins of around 80% per token before the latest reductions, potentially giving them room to lower prices while remaining profitable on inference.
The changes have also affected how developers manage AI spending.
Woolf said GPT-6.1 Sol’s pricing reduced his effective usage costs by approximately 75%. He subsequently lowered his Codex usage quota to one-fifth of its previous level while keeping his overall spending roughly unchanged.
At the same time, OpenAI reduced the usage allowance for its $200-per-month Codex tier by 50%, citing increasingly efficient models.
AI Model Prices Could Fall Further
Woolf believes AI model prices could continue declining as long as competing inference providers maintain similar performance and pricing while allowing customers to switch services with minimal difficulty.
Meanwhile, OpenAI and Anthropic are trying to strengthen customer relationships through enterprise agreements and personal AI agents, including OpenAI Dots and Claude Duet.
However, growing competition could eventually push prices closer to the actual cost of running AI models.
If current trends continue, developers could benefit from further reductions in AI usage costs, while major AI companies face increasing pressure to improve performance and maintain profitability.
The post OpenAI and Anthropic Slashed AI Model Prices by Up to 80% appeared first on ProPakistani.
