/
Navigation
Chronicles
Browse all articles
Explore
Semantic exploration
Research
Entity momentum
Nexus
Correlations & relationships
Story Arc
Topic evolution
Drift Map
Semantic trajectory animation
Posts
Analysis & commentary
Pulse API
Tech news intelligence API
Browse
Entities
Companies, people, products, technologies
Domains
Browse by publication source
Handles
Browse by social media handle
Detection
Concept Search
Semantic similarity search
High Impact Stories
Top coverage by position
Sentiment Analysis
Positive/negative coverage
Anomaly Detection
Unusual coverage patterns
Analysis
Rivalry Report
Compare two entities head-to-head
Semantic Pivots
Narrative discontinuities
Crisis Response
Event recovery patterns
Connected
Search: /
Command: ⌘K
Embeddings: large
TEXXR

Chronicles

The story behind the story

days · browse · Enter similar · o open

A look at OpenAI's open-source agent harness that now powers Codex and ChatGPT Work, as the company works to optimize the harness to cut runaway token usage

The unsung hero behind OpenAI's 2026 transformation is a technology that rarely gets mentioned, and that most of its 1 billion monthly users have never even heard of.

The Deep View Jason Hiner

Context & Ripple Effects

Codex began as a software-engineering agent and later became the shared-agent foundation for ChatGPT workspaces. OpenAI’s own reported rollout of Codex-powered workspace agents made the harness a common layer across both products.

That shared layer is now operating at materially greater reach: OpenAI recently said Codex and ChatGPT Work had reached 10 million users, while internal Codex use had also expanded sharply in OpenAI’s own workforce. Token control therefore matters as an operating constraint, not merely an engineering refinement.

First-order effects

  • Optimizing the open-source harness directly changes how Codex and ChatGPT Work execute agent tasks, with the immediate goal of curbing runaway token consumption across both products.
  • OpenAI gains a single leverage point for improving the efficiency of two growing agent surfaces rather than tuning each application independently.

Second-order effects

  • Lower token use can improve the economics of serving longer or more complex agent workflows, reducing pressure on product teams to constrain usage solely to contain inference costs.
  • Because the harness is open source, its optimization work can become a reference point for developers building compatible agent workflows, extending its influence beyond OpenAI’s own interfaces.

Third-order effects

  • The episode points to agent runtimes becoming a distinct competitive layer: model quality alone is insufficient when orchestration determines whether autonomous workflows remain economically usable.
  • If adoption continues to broaden from coding into shared workspaces, governance and cost controls in the runtime will increasingly shape which agent products can scale reliably.

The trend: AI companies are shifting competition from standalone models toward efficient, governed runtimes that can support high-volume agent work across multiple product surfaces.