OpenAI readies GPT-5 for August, promising one model for everything

OpenAI prepares to release GPT-5 in early August, combining advanced reasoning with multimodal functions in a single model that represents the company’s most powerful artificial intelligence (AI) system to date.

Multiple sources from Reuters and Axios confirm the next-generation model will debut within weeks. The release represents a significant shift from the company’s current approach of maintaining separate model lines.

GPT-5 merges OpenAI’s reasoning-focused “o-series” models with the multimodal capabilities of its traditional GPT lineup. This unification eliminates the need for users to switch between different models for various tasks.

“The breakthrough of reasoning in the O-series and the breakthroughs in multi-modality in the GPT-series will be unified, and that will be GPT-5,” said Romain Huet, OpenAI’s Head of Developer Experience.

GPT-5 reportedly demonstrates particular strength in software engineering tasks. Navigation of complex, legacy codebases represents an area where previous OpenAI models struggled against competitors like Anthropic’s Claude.

CEO Sam Altman has begun publicly discussing his experience with the unreleased version. During a Theo Von podcast appearance, he described testing GPT-5 on a question he couldn’t answer himself.

“I put it in the model, this is GPT-5, and it answered it perfectly,” Altman said on the Theo Von podcast. “I felt useless relative to the AI.”

The launch timeline faced multiple delays from an original late May target. Additional safety testing and infrastructure preparation pushed the release into August, according to sources close to the development process.

Microsoft engineers have been preparing server capacity for months. The tech giant is already testing internal builds of Copilot and Microsoft 365 Copilot with GPT-5 integration.

OpenAI plans to offer three versions at launch. The full model will be available through ChatGPT and the company’s API. A “mini” version will provide faster processing for simpler tasks. A compact “nano” model will serve API customers exclusively.

The model may temporarily use “routing” technology at launch to automatically direct queries to specialized reasoning capabilities when needed. This approach allows GPT-5 to avoid unnecessary computational overhead for straightforward requests.

GPT-5 access will likely require a ChatGPT Plus subscription at $20 monthly. Enterprise and Team users should receive priority access, following OpenAI’s established rollout pattern.

The launch occurs amid intensifying AI competition. Recent administration statements emphasize that America must “do whatever it takes” to maintain technological leadership over China in artificial intelligence development.

OpenAI’s first open-weights model since 2019 may also arrive before or after GPT-5, depending on final scheduling decisions.