OpenAI announces 80% price drop for o3, it’s most powerful reasoning model

Good news, artificial intelligence developers!

Openai has announced a significant price reduction on O3, the Great Language Model (LMM), The costs of reduction by 80 % are huge for each of the distinctive symbols of inputs and outputs.

(Mention the symbols are the individual digital strings that LLMS uses to represent words, phrases, sports chains, coding, and other content. They are representations of the semantic construction that the model teaches through training, and in its essence, the developers who receive them in order to obtain protected applications or their external court.

The update places the model as an easy option for developers looking for advanced thinking capabilities, and put Openai in a more direct rate competition with competing models such as Gueini 2.5 Pro from Google Deepmind and Claude OPUS 4 from Suite Anthropco and Deepseek.

He announced by altman himself on x

Samtman, CEO of Openai, confirmed the change in a post on X highlighting that the new pricing aims to encourage the wider experiences, and writing: “We have dropped the price of O3 by 80 % !!

The cost of using O3 now is $ 2 per million input symbols and $ 8 per million output symbols, with an additional discount of $ 0.50 per million symbols when the user enters “hidden” information, or is stored and similar to what they have provided before.

This is a significant discount from the previous rates of $ 10 (input) and $ 40 (output), indicated by Openai Noam Brown Researcher on X.

Ray Fernando, an early developer and adoption, celebrated a decrease in pricing in the book “LFG!” A shortcut to “Let’s go ridiculous!”

Feelings reflect an increased enthusiasm among builders who are looking to expand their projects without the costs of reaching the banned model.

Comparison of the price with other competition LLMS

Pricing modification comes at a time when artificial intelligence providers are more competing for both performance and the ability to withstand costs. The comparison with other artificial intelligence models shows how important this step is:

• Gemini 2.5 Pro Preview,, Which was developed by Google DeepMind, ranging from $ 1.25 and $ 2.50 for inputs depending on the claim volume, and $ 10 to $ 15 to output. Although its integration with Google Search provides additional functions, this service carries its own cost – for free for the first 1500 requests per day, then $ 35 per thousand requests.
• Claude Obus 4,, Marketing by Antarbur as an improved model for complex tasks, is the most expensive group, as it receives $ 15 per million input symbols and $ 75 for production. Quickly storage reading and writing services come at $ 1.50 and $ 18.75, respectively, although users can open a 50 % discount with payment processing.
• Deepseek models, especially DeepSeek -resons and Deepseek-Cat, undermine a lot of market with low aggressive prices. The distinctive symbols of inputs range from $ 0.07 to $ 0.55 depending on the temporary storage and time today, while the output ranges from $ 1.10 to $ 2.19. Discounts during peak hours outside the prices drop further, to up to $ 0.035 for temporary stored inputs.

model	entrance	Inputs stored temporarily	Output	Discount notes
Openai o3	$ 2.00 (a decrease from 10.00 dollars)	0.50 dollars	8.00 dollars (decrease from $ 40.00)	Flexible treatment: $ 5 / $ 20
Gemini 2.5 Pro	$ 1.25 – $ 2.50	0.31 dollars – 0.625 dollars	$ 10.00 – 15.00 dollars	A higher rate applies to claims> 200 kg symbols
Claude Obus 4	$ 15.00	$ 1.50 (read) / 18.75 dollars (write)	$ 75.00	50 % discount with payment processing
Deepseek-Cat	0.07 dollars (hit) $ 0.27 (Miss)	–	$ 1.10	50 % discount during peak hours
Deepseek-RESonser	0.14 dollars (hit) $ 0.55 (Miss)	–	$ 2.19	75 % discount during peak hours

In addition, the AI ​​Group’s II’s IIS IIS Three O3 third -party operation through the standard test set at various tasks, and found that the cost of $ 390 to complete it all, compared to $ 971 for Gemini 2.5 Pro and $ 342 Cleude 4 Sonnet.

Narrowing the cost against the smart gap of developers

Not only narrows the OpenAI pricing movement with low -cost models such as Deepseek, but also put down the declining pressure on high -rate offers such as Claude Opus and Gemini Pro.

Unlike Clade or Gemini, Openai’s O3 also offers a flexible position for simultaneous treatment that receives $ 5 to inputs and $ 20 to take out per million symbols, giving developers more control of account cost and cumin depending on the type of work burden.

O3 is currently available through Openai API and Playground. Users who have low balances up to a few dollars can explore the full capabilities of the model, allowing initial models and posting with fewer financial barriers.

This can be particularly beneficial to startups, research teams and individual developers who have previously found up to upper models to reach the cost.

By reducing the cost of the most advanced thinking model, Openai refers to a wider direction in the area of ​​artificial intelligence: Excellent performance has become more at a reasonable prices, and developers now extend an increasing number of economically developed options.