OpenAI’s aggressive pricing for GPT-5 sparks fears of a looming AI model price war

OpenAI has once again disrupted the AI industry by launching GPT-5 at a remarkably low price, sparking early speculation of a broader price war in the large language model market. Just days after releasing two new open-source models, the company unveiled GPT-5, which CEO Sam Altman hailed as “the best model in the world.” While benchmarks show GPT-5 performs only slightly better than leading models from Anthropic, Google DeepMind, and xAI in certain areas—and trails in others—its overall versatility, especially in coding tasks, has drawn strong praise. The pricing is where GPT-5 truly stands out. The top-tier API charges $1.25 per million input tokens and $10 per million output tokens, with an additional $0.125 per million for cached input. This positions it competitively with Google’s Gemini 2.5 Pro Basic, which has similar rates but imposes higher fees for heavy usage beyond 200,000 prompts. OpenAI’s model, by contrast, maintains consistent pricing even at scale. More significantly, GPT-5 undercuts Anthropic’s Claude Opus 4.1, which starts at $15 per million input tokens and $75 per million output tokens. Although Anthropic offers discounts for prompt caching and batch processing, OpenAI’s base pricing remains far more attractive for many users. Developers with early access have lauded the value. Simon Willison, featured in OpenAI’s launch video, called the pricing “aggressively competitive.” Matt Shumer, CEO of OthersideAI and creator of HyperWrite, noted that GPT-5 is cheaper than GPT-4o, calling it a win for “intelligence per dollar.” On social platforms like X and Hacker News, users have described OpenAI’s pricing as a “pricing killer,” signaling strong market reaction. This aggressive pricing could pressure competitors to respond. Google previously undercut OpenAI on cost, and Meta has shown interest in lowering prices as part of its broader AI strategy. If others follow suit, a full-blown LLM price war could emerge—something many in the developer and startup communities have long anticipated. Such a shift would be welcome. Many AI-powered tools, especially those focused on “vibe-coding” and other high-usage applications, face unstable economics due to unpredictable and rising API costs. As TechCrunch’s Marina Temkin has reported, these costs remain a major hurdle for startups building on top of foundational models. Despite massive infrastructure investments—OpenAI’s $30 billion annual contract with Oracle, Meta’s projected $72 billion AI spend in 2025, and Alphabet’s $85 billion capital allocation—costs have historically trended upward. Yet OpenAI’s dual move this week—releasing open-source models and launching GPT-5 at low prices—has thrown a challenge into the market. While it may be too early for startups to celebrate, OpenAI has clearly signaled that pricing innovation is accelerating. The real test now is whether others will match or exceed it.