Product Launchagentsllmdeveloper toolsopenai
GPT-5 Codex Builds Its Own Codebase
9.0
OpenAI is deploying GPT-5 Codex as a self-improving coding agent, with Ars Technica and company posts reporting the model now builds the majority of its own code. The update, rolled into GitHub Copilot and GPT-5.2 previews, enables long-running workflows exceeding seven hours, supports 400K-token contexts, and hits benchmarks like 74.5% on SWE-bench Verified.
Key Points
- 1Establishes recursive development: Codex generates and refines most of its own codebase for agentic improvements.
- 2Reduces human labeling and speeds iterations, achieving benchmarks like 74.5% on SWE-bench Verified.
- 3Enables long-running workflows over seven hours and 400K-token context, boosting codebase automation capabilities.
Scoring Rationale
High novelty and industry-wide scope, but practical adoption details and long-term safety controls remain limited.
Sources
Public references used for this report.
Practice with real Logistics & Shipping data
90 SQL & Python problems · 15 industry datasets
Used by DS/ML engineers at top companies
High-Value Overnight OrdersEasyDelivered International ShipmentsMediumOn-Time Delivery Rate by CarrierHard
250 free problems · No credit card
See all Logistics & Shipping problems