Openai surprised the tech industry for the second time this week by launching its latest flagship model, the GPT-5, just a few days after releasing two new, freely available models under the open source license.
Openai CEO Sam Altman called the GPT-5 “the best model in the world.” It could be pride or exaggeration, as TechCrunch’s Maxwell Zeff reports that the GPT-5 is slightly ahead of humanity, Google Deepmind, and other major AI models from Xai.
Still, it is a model that works well for a variety of uses, especially for coding. And, as Altman pointed out, one area that it definitely competes with is price. “I’m very pleased with the prices we can offer!” he tweeted.
The top-level GPT-5 API is $1.25 per million tokens per input and $10 per million tokens for output (and $0.125 per million tokens for cache input). This price reflects Google’s Gemini 2.5 Pro Basic Subscription. This is also popular for coding-related tasks. However, Google will charge more if the input/output exceeds the 200,000 prompt heavy threshold.
However, Openai actually hides humanity’s Claude Opus 4.1. This starts at $15 per input token and $75 per output token. (However, humanity offers a big discount on prompt cache and batch processing prompts. It provides both the storage/reuse of prompts and the handling of multiple requests together.)
The Anthropic model is extremely popular among programmers as an option within the popular coding assistant cursor and as a power source for its own assistant, Claude Code. (Note that Cursor offered GPT-5 as an option a few minutes after its announcement.)
Developers who have early access to GPT-5 are advertising prices. Simon Willison, one of the developers featured in Openai’s launch video, wrote in his review: “Prices are actively competitive with other providers.”
TechCrunch Events
San Francisco
|
October 27th-29th, 2025
However, the GPT-5 is also a competitive price for the GPT-4O. Other IDEAI co-founder and CEO Matt Shumer (manufactured by HyperWrite) said the GPT-5 was “cheaper than the GPT-4O, which is great. Intelligence continues to increase.”
Part of the X is called Openai’s price for the model “A Pricing Killer”, and others on Hacker News offer similar praise.
Do you follow competitors like your competitors? Will Google previously defeat Openai on pricing? If so, we may be witnessing the beginning of the much-anticipated LLM price war.
There is no doubt that a price war will be welcomed. For example, the underlying economics of Vibe-Coding Tool providers are rather unstable due to the high and unpredictable fees that require you to pay the model maker, as reported by TechCrunch’s Marina Temkin. There are also countless startups built on top of AI models.
Silicon Valley hopes that the price-to-performance ratio of LLM will ultimately improve and inference costs will improve. However, such equalization seemed to be apart for years as the tech industry invested hundreds of billions to build data centers and infrastructure to support increased demand for AI.
Openai itself has a $30 billion annual contract with Oracle when it reached just $10 billion, which achieved recurring revenue of $10 billion a year. Meanwhile, META is planning to spend up to $72 billion on AI infrastructure in 2025, while Alphabet set $85 billion in 2025 based on AI needs. In the face of such enormous costs, costs usually go in one way.
Given these investments, it may be too early to please Openai’s move to only pricing for startups to consider rising model API invoices.
However, this week, Openai tossed the Gauntlet, putting pressure on the price not just once but twice. Check if others will follow.
Source link
