LLM Reference

SWE-1 Models by Cognition

CognitionProprietaryCodingAgents
1 model2026

Details

ResearcherCognition
LicenseProprietary
Commercial useCommercial use: conditional
Models1
Released2026

Capabilities

ReasoningAll models
Function CallingAll models
Tool UseAll models
Code ExecutionAll models

Links

Website

About

Cognition's SWE-1 coding model family for agentic software-engineering work, available through Devin product surfaces and optimized for long-horizon asynchronous coding tasks. The current first-party documented version is SWE-1.7, launched July 8, 2026, trained from the Kimi K2.7-Code base with additional RL.

Current Variants

Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.

1 in view
SWE-1Current

Use when the workload needs coding, reasoning, and tool use.

2026-07codingreasoningtool use

Release Timeline

1 release group
2026-07
1 current
SWE-1
codingreasoningtool use
Current

Specifications(1 models)

SWE-1 model specifications comparison
ModelReleasedReasoningFn CallingTool UseCode Exec
SWE-12026-07YesYesYesYes

Available From(1 provider)

Frequently Asked Questions

What is SWE-1 used for?
SWE-1 is used for coding, agents, and reasoning. The family description and listed model capabilities point to those workloads as the best fit.
How does SWE-1 compare to Agents-A1?
SWE-1 by Cognition is strongest where you need coding, while Agents-A1 by Intern Science is the closest related family to check for agents. SWE-1 has 1 listed variant, while Agents-A1 reaches up to 262k context, so compare the specs and pricing tables before choosing a production model.
Which SWE-1 model should I use?
If price is the main constraint, use the pricing table first because SWE-1 does not have complete provider pricing in the local data. For the most capable/latest local choice, evaluate SWE-1 with reasoning, tool use, and function calling.