SWE-2
5.0 out of 5 starsAverage Ratings 0 Ratings
Average Ratings 1 Rating
Description
MAI Code 2 is an advanced artificial intelligence coding model designed to assist developers with software engineering, application development, and complex code maintenance tasks. Developed by Microsoft AI, the model focuses on practical enterprise programming workloads such as implementing features, debugging applications, refactoring code, and preparing changes for production. Its architecture combines a decoder-only transformer with sparse Mixture-of-Experts layers and an integrated vision encoder for processing both text and images. The model provides a context window of approximately 1.05 million tokens, allowing it to examine extensive source code, documentation, and related development materials. MAI Code 2 supports repository-wide reasoning and multi-file modifications, helping developers implement coordinated changes across interconnected software components. Its frontend development capabilities include generating user interfaces, dashboards, data visualizations, and interactive applications from developer instructions. Microsoft reports that the model delivers 98% of GPT-6.1 Sol's coding benchmark performance at 40% of the cost and 96% of Claude Opus 5.5's performance at 5% of the cost under its published evaluation conditions. The model is available in GitHub Copilot, with additional distribution through developer platforms planned. By combining long-context code understanding, automated implementation, debugging assistance, and visual programming capabilities, MAI Code 2 helps engineering teams improve development productivity while managing AI inference costs.
Description
SWE-2 is a software engineering model from Cognition built for agentic coding tasks that require strong performance at lower computational and monetary cost. It is post-trained from the Kimi K3 base model and extends Cognition’s earlier SWE-1.7 training approach with a new reinforcement learning method for jointly optimizing multiple reasoning-effort settings. Medium, high, and maximum effort modes provide different tradeoffs between speed, cost, exploration, and verification depending on task complexity. The model is trained to inspect only the parts of a codebase that are likely to matter, helping it reach implementation faster and reduce unnecessary exploration. SWE-2 can generate and modify code, run tests, analyze repositories, work through terminal tasks, and verify whether implementations satisfy user requirements. Cognition also reports improvements in end-to-end test creation, regression detection, instruction following, and re-deriving conclusions when challenged. Its training process incorporates cost-aware rewards, length-weighted reward baselines, expanded reinforcement learning environments, and hardened verifiers intended to improve both efficiency and reliability. SWE-2 is positioned as a cost-efficient alternative to larger frontier coding models while remaining competitive on software engineering benchmarks such as FrontierCode, DeepSWE, and Terminal-Bench. The model is available in Devin Desktop and Devin CLI and is being introduced to additional Cognition products including Devin Web and Fusion.
API Access
Has API
No
API Access
Has API
No
Integrations
.NET
Yes
C
No
C#
No
C++
No
CSS
No
Devin
No
Devin Desktop
No
Go
No
JSON
No
Microsoft Azure
Yes
Integrations
.NET
Yes
C
Yes
C#
Yes
C++
Yes
CSS
Yes
Devin
Yes
Devin Desktop
Yes
Go
Yes
JSON
Yes
Microsoft Azure
No
Pricing Details
$0.50 per 1M tokens (input)
$0.50 per 1 million input tokens and $2.50 per 1 million output tokens (with cached input reads at $0.025 per 1M tokens)
Free Trial
No
Free Version
No
Pricing Details
$20/month
Free Trial
No
Free Version
Yes
Deployment
Web-Based
Yes
On-Premises
No
iPhone App
No
iPad App
No
Android App
No
Windows
No
Mac
No
Linux
No
Chromebook
No
Deployment
Web-Based
Yes
On-Premises
No
iPhone App
No
iPad App
No
Android App
No
Windows
No
Mac
No
Linux
No
Chromebook
No
Customer Support
Business Hours
No
Live Rep (24/7)
No
Online Support
Yes
Customer Support
Business Hours
No
Live Rep (24/7)
No
Online Support
Yes
Types of Training
Training Docs
Yes
Webinars
No
Live Training (Online)
No
In Person
No
Types of Training
Training Docs
Yes
Webinars
No
Live Training (Online)
No
In Person
No
Vendor Details
Company Name
Microsoft AI
Founded
2024
Country
United States
Website
microsoft.ai/
Vendor Details
Company Name
Cognition
Founded
2023
Country
United States
Website
cognition.com