OpenAI Chief Executive Officer Sam Altman said on an internal company podcast that the next flagship language model, GPT-5, is scheduled for release later this summer, subject to final internal tests. The model is billed as “significantly better” than GPT-4 and aims to reduce hallucinations while offering a single interface for a broad set of tasks. Developer-relations executive Romain Huet added that GPT-5 will combine the advanced logical reasoning introduced in the recent O-series with the multimodal capabilities of the main GPT line, creating a unified architecture meant to handle text, images and other inputs without requiring users to switch models. Rival research outfit xAI is preparing its own upgrade, Grok 4. Developers close to the project describe the forthcoming release as a bigger advance than the step from Grok 2 to Grok 3, with plans for a native image generator, an expanded context window and a command-line coding agent intended to compete with similar tools from Anthropic and others. The parallel roadmaps underscore intensifying competition among leading artificial-intelligence labs as they race to bring more powerful, general-purpose models to market in the second half of 2025.