Agent-harness co-design
Cognition trained SWE-1.5 with reinforcement learning inside its Cascade coding environment rather than treating the model and agent tools as separate systems.
Historical tool archive
SWE-1.5: historical overview
SWE-1.5 was Cognition's speed-focused agentic coding model for Windsurf; it has since been replaced by SWE-1.6 and the current SWE-1.7 family.
No longer active
Archive
What SWE-1.5 was
Cognition released SWE-1.5 in October 2025 as a software-engineering model co-optimized with the Cascade agent harness. Served through Cerebras at up to 950 tokens per second, it emphasized rapid codebase exploration, configuration edits and end-to-end coding tasks.
SWE-1.5 is no longer part of Cognition's current model lineup. SWE-1.6 improved its agent behavior and benchmark performance, and SWE-1.7 is now the latest in-house model available across Devin Web, Desktop and CLI.
Past capabilities
What SWE-1.5 offered
Fast inference
The model was served through Cerebras at up to 950 tokens per second, with surrounding lint and command pipelines optimized to reduce delay.
Software-engineering focus
Its training environments covered codebase navigation, configuration work, full-stack changes, automated tests and browser-based grading.
Windsurf integration
SWE-1.5 launched inside Windsurf and also powered the beta Codemaps experience.
Availability
Why this listing is historical
SWE-1.5 did not have a standalone public subscription and is no longer offered in Cognition's current model selector. Current Devin plans and model usage have separate pricing.
- No longer offered in Cognition's current model lineup
- Cognition reported that SWE-1.6 improved SWE-Bench Pro performance by more than 10 percent over SWE-1.5
- Later company testing found SWE-1.5 was not strong enough as the primary model in some multi-agent workflows
Status checked . Review the historical source ↗
Current options
Active SWE-1.5 alternatives
Coding
SWE-1.7
Use SWE-1.7 for Cognition's current, more capable in-house coding model.
Explore SWE-1.7 →Coding
SWE-1.7 (successor to SWE-1.6)
SWE-1.6 directly succeeded SWE-1.5 and remains listed as a previous-generation option.
Explore SWE-1.7 (successor to SWE-1.6) →Coding
Composer 2.5
Consider Composer 2.5 for Cursor's current in-house model optimized for longer agent sessions.
Explore Composer 2.5 →Coding
Codex
Consider Codex for an actively maintained coding-agent workflow across local and cloud environments.
Explore Codex →Questions
SWE-1.5 FAQs
Is SWE-1.5 still available?
No. Cognition's current model documentation lists SWE-1.7 and SWE-1.6 but no longer lists SWE-1.5.
What replaced SWE-1.5?
SWE-1.6 directly followed it, and SWE-1.7 is now Cognition's latest in-house software-engineering model.
Was SWE-1.5 a standalone API?
It launched as a model inside Windsurf, not as a separately priced general-purpose API.
Similar Coding Tools
View all Coding AI tools →
.png)
AI training for the future of work.
Get access to all our AI courses, hundreds of real-world AI use cases, live expert-led workshops, an exclusive network of AI early adopters, and more.
AI Courses
Get unlimited access to all of our current & upcoming industry-specific AI courses for the duration of your subscription.
Daily Guides
To keep up with the rapid pace of AI, our team publishes AI implementation guides daily. Our library contains 300+ practical use cases to automate real-world work.
Workshops
Join weekly, live, interactive sessions with industry leaders who are at the forefront of AI for hands-on implementation guidance and exclusive insights.
Community
Network with an exclusive community of AI-first professionals who are working smarter with AI. Learn how early adopters are using AI in their work and businesses.