Details
- OpenAI has launched GPT-5 via its API platform as of August 28, 2025, specifically targeting developers and offering leading-edge capabilities in coding and agentic tasks, with benchmark scores of 74.9% on SWE-bench Verified and 88% on Aider Polyglot.
- The release includes three distinct model sizes—GPT-5, GPT-5 Mini, and GPT-5 Nano—with a pricing range from $0.05 to $10 per million tokens, and introduces new API features such as verbosity controls, reasoning effort parameters, and support for custom tool integration.
- Designed to outperform its predecessors in tasks like frontend development, GPT-5 surpasses the o3 model in 70% of head-to-head tests, while achieving these results with 22% fewer output tokens and 45% fewer tool calls, underscoring both its performance and efficiency.
- The GPT-5 API version focuses on maximizing developer utility by offering an expanded context window up to 400,000 tokens and more robust tool-calling abilities for handling complex, multi-step tasks, distinguishing it from the ChatGPT release.
- Following the initial August 7 debut on ChatGPT and Microsoft platforms, the API has quickly attracted early users—including Cursor, Windsurf, Vercel, and GitHub Copilot—who have reported notable gains in coding assistance and the automation of agentic workflows.
Impact
OpenAI’s GPT-5 API launch intensifies competition in the AI developer tools landscape, challenging rivals like Claude and Gemini with its tiered pricing and high performance. The new features and scalable options cater to a broad spectrum of development needs, signaling a shift toward highly autonomous, AI-driven coding practices that could redefine industry standards in the coming year.