OpenAI claims GPT-5 model boosts ChatGPT to 'PhD level'. GPT-5 is now available in our API platform, delivering unmatched performance in coding, reasoning, and agent-based workflows. Built to be a true coding collaborator, it sets new benchmarks in real-world software engineering and multi-tool orchestration.
Key Highlights:
- SWE-bench Verified: 74.9% (highest recorded)
- Aider Polyglot: 88% multi-language code editing
- Outperforms OpenAI o3 in frontend development 70% of the time
- Handles dozens of sequential or parallel tool calls with high reliability
- Provides configurable verbosity and reasoning effort settings
- Supports custom tools with plaintext or regex inputs
- 400K context window for large-scale projects
Built for:
- Frontend engineering with higher aesthetic and functional accuracy
- Deep codebase reasoning and system explanation
- Automated, multi-language refactoring
- Long-running autonomous agents
Benchmarks:
SWE-bench Verified: GPT-5 74.9% vs o3 69.1%
Aider Polyglot: GPT-5 88% vs o3 79.6%
τ2-bench Telecom: GPT-5 96.7% vs o3 58%
Models & Pricing:
gpt-5, gpt-5-mini, gpt-5-nano
Input: $0.05 to $1.25 / 1M tokens
Output: $0.40 to $10 / 1M tokens
Available now in the OpenAI API, Codex CLI, Microsoft 365 Copilot, GitHub Copilot, and Azure AI Foundry.
GPT-5 is the new standard for intelligent, reliable, and scalable software development.
hashtag#GPT5 hashtag#gpt5 hashtag#gpt-5 hashtag#openai hashtag#chatgpt
Top comments (0)