OpenAI open-sources Codex Harness
09/30/2026 — 10/02, 23:54·1 sources·1 reports
Story overview
A report published by Leiphone on September 30 says OpenAI officially open-sourced Codex Harness on August 19. According to the report, the runtime methodology behind Harness had until then been held by only a small number of top Agent products, and the release opens it up to a wider group of developers and enterprises. The article sums up the structure of an Agent with a formula: Agent = Model + Harness, and says developers can now build on top of a ready-made Agent Runtime instead of assembling one themselves.
Developers who have worked on Agent Harness for a long time reacted to the news with what the report describes as excitement, since capabilities previously confined to a handful of leading Agent products are now available to more developers and companies. The article argues that as best practices such as Codex are open-sourced, Harness is shifting from private property to a public resource, and its scarcity is falling quickly. Technical leads that many companies built through Harness Engineering will, in this view, be learned and copied across the industry sooner than before. The report frames this as the reason OpenAI finally "opened up" this time, and asks in its headline what is left of the moat.
The report does not specify the exact scope of the Codex Harness release, its license terms, or any follow-up version plans, and it offers no competitive or commercial data beyond the developer reaction described.
AI-generated from 1 reports · updated 2 hours ago
Latest turnOpenAI open-sourced Codex Harness on August 19, releasing an agent runtime that only a handful of top agent products had kept private. Harness is the layer captured by the formula Agent = Model + Harness, so developers can now build on a ready-made Agent Runtime. The piece argues that leads built on Harness Engineering will be copied across the industry faster.

Reports on this story headlines open the original
OpenAI open-sourced Codex Harness on August 19, releasing an agent runtime that only a handful of top agent products had kept private. Harness is the layer captured by the formula Agent = Model + Harness, so developers can now build on a ready-made Agent Runtime. The piece argues that leads built on Harness Engineering will be copied across the industry faster.
Other stories people are talking about
- 452NewNVIDIA launches 64GB DGX Spark desktop AI computer at $4,9995 sources
- 278Google releases Gemini 4 Argon as next-gen frontier AI model9 sources
- 232NewSurgeGemini 4 Argon tops the Arena AI leaderboard5 sources
- 232SurgeMicrosoft AI releases MAI-Transcribe-2-Streaming real-time transcription model3 sources
- 197NewAnthropic launches Claude Frontier Academy with $100M2 sources
- 185NewHugging Face open-sources AstaBrief for fast report generation2 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion
