Warp said its newly open-sourced terminal now underpins “Open Agentic Development,” a workflow where human developers supervise agents that plan work, write code, test changes and open pull requests.
GPT-5.5 is central to those workflows: Warp said the model handled larger reasoning tasks and used 30% fewer tokens per agentic coding task than GPT-5.4 in internal benchmarks.
Oz, Warp’s orchestration platform, coordinates those agents across local and cloud environments with shared memory, live monitoring, recurring jobs and model routing for harder coding and reasoning tasks.
Warp said agents already co-create about 90% of its pull requests; the company now has nearly 1 million developers and is used by more than 56% of Fortune 500 companies.
The push comes as Warp bets software development will shift from individual coding assistants to persistent fleets of agents, a strategy it says helped drive ARR up 35x last year.
As AI generates 90% of its code, how does Warp prevent catastrophic security flaws before they are deployed?
With AI agents handling coding, what is the new essential skill for software developers to remain valuable?
Can an open-source AI model truly out-innovate heavily funded, closed-source competitors in the long run?
Inside Warp’s Agent-First Revolution: Open-Sourcing, Oz Platform, and the Future of AI-Native Software Development
Overview
In April 2026, Warp made a major move by open-sourcing its client, a decision rooted in its original vision and driven by the rapid rise of AI agents. Recognizing that staying closed would be a missed opportunity, Warp shifted to an agent-first approach, aiming to build a more collaborative and expansive ecosystem. By opening its client, Warp seeks to harness the collective intelligence of the developer community and align with the growing importance of AI agents. This strategic shift positions Warp to lead in the evolving landscape of AI-native development, fostering innovation and broader participation.