The latest AI news and updates of Cognition — Agent lab building Devin, a collaborative AI teammate for software engineering teams. Covering Cognition's latest product updates, company news, and launches from the past 90 days.
Cognition23h agoCognition upgraded Devin Security to perform deep vulnerability scans using Claude Mythos and other frontier cyber models. The platform builds custom threat models to prioritize findings, then reproduces and patches vulnerabilities in parallel virtual machines. Cognition also committed hundreds of millions of dollars in credits to provide enterprise customers with free scans and remediation support.
CognitionOct 23Cognition released a refreshed Devin Desktop app featuring an Agent Command Center as the default IDE surface. The update introduces Spaces for shared agent context and support for the Agent Client Protocol to run third-party agents. The app remains backwards-compatible with Windsurf, is free to download, and integrates with existing ChatGPT subscriptions for model usage.
CognitionOct 22Cognition open-sourced the Vulnerability Fixes plugin to automate remediation for recent PHP and Node.js remote code execution flaws. The plugin performs a repository inventory, applies verified patches, runs before-and-after tests, and opens focused pull requests. It installs in Devin via the marketplace or in other coding agents using npx skills. Devin Enterprise customers receive additional remediation support and credits.
CognitionOct 20Cognition launched Devin Fusion, a multi-model harness that pairs a frontier lead model with a cost-effective sidekick for software engineering tasks. Configured with Claude Opus 5.5 and GPT-6 Luna, the harness scored 68 on the Artificial Analysis Coding Agent Index. Cognition reports this configuration achieves a 31% lower cost per task compared to Claude Code with Opus 5.5.
CognitionOct 19Cognition launched efficiency-based cost controls for Devin, featuring an Efficiency Score, Devin Coach, and auto-approvals. The Efficiency Score grades usage from 0–100 based on productive engineering hours per ACU. Devin Coach provides prompt feedback, while auto-approvals automatically raise limits for Healthy or Satisfactory engineers up to an admin-set ceiling. These features are available today for Devin Enterprise customers.
CognitionOct 15Cognition launched Devin Oncall, an autonomous agent that investigates production alerts and bug reports to diagnose and resolve incidents. The agent already handles thousands of daily alerts for large companies and integrates out of the box with Datadog, Dynatrace, and Sentry. It automates triage and remediation, reducing the manual burden of on-call rotations.
CognitionOct 13Cognition has integrated Google's Gemini 4 Argon into Devin Desktop and Devin CLI. The model scores 60.5% on the FrontierCode 1.1 benchmark, marking the highest performance for any Gemini model on the leaderboard. Devin users receive a 75% discount on the model through October 27, 2026.
CognitionOct 8Cognition has integrated Anthropic's Claude Haiku 5.5 into Devin. The model scores 58.4% on the FrontierCode 1.1 benchmark, outperforming Sonnet 5 at about one-eighth the cost per task. It now serves as a sidekick model within Devin Fusion, pairing with Opus 5.5 as the lead in Devin Desktop and CLI environments.
CognitionOct 5Cognition launched Dreaming for its Devin AI software engineer, enabling the agent to build a memory graph of user preferences and project conventions across sessions. At night, Devin self-improves this memory by consolidating notes and removing stale records. Cognition also released the Agent Memory Repo, an open-source standard that stores memory records as Git-versioned files.
CognitionOct 1Cognition is the first customer running NVIDIA's Vera Rubin platform, powered by CoreWeave. Measured on its SWE-2 coding model, the hardware delivers about 4.8x more token throughput than the previous GB200 architecture at the same decode speed. This deployment marks the latest generation of infrastructure for Cognition's autonomous software engineering agents.
CognitionSep 30Cognition’s Devin AI software engineer now serves as the first responder for software bugs and alerts at AI-native law firm Crosby. Devin investigates 20 to 40 issues daily, reducing Sentry alert noise by 50%. This automation allows Crosby to maintain a single-engineer on-call rotation even as its legal team has grown fivefold.
CognitionSep 29Cognition launched Devin for MongoDB Modernizations, integrating its AI software engineer into MongoDB’s Application Modernization Platform. Devin autonomously handles legacy code rewriting and business logic, while AMP manages data movement and validation. This joint approach automates the migration process, accelerating the transition of legacy applications to MongoDB Atlas from years to months.
CognitionSep 29Cognition now integrates ChatGPT Plus and Pro subscriptions into Devin. OpenAI model usage across Devin Cloud, Desktop, and CLI draws from the user's existing ChatGPT quota. Settings allow usage limits for Devin, and the platform prioritizes OpenAI models when a subscription is connected. This integration is available on Devin Pro, Max, and Teams plans.
CognitionSep 29Cognition has integrated OpenAI's GPT-6.1 Sol into Devin Desktop and CLI. On the FrontierCode 1.1 benchmark, the model scores 60.4%, matching GPT-6 Sol's performance at a fraction of the cost. At medium reasoning effort, it costs $0.31 per task, while its low-effort setting achieves a 58.1% score for $0.21 per task, the highest for any model under $0.30.
CognitionSep 28Cognition added Anthropic's Claude Sonnet 5.5 to Devin Desktop and Devin CLI. The model scores 64.4% on FrontierCode 1.1 Main, improving on Sonnet 5's 56.2% and surpassing Fable 5.1 at extra high reasoning effort. Anthropic reports over 30% faster output generation and up to 30% lower cost per task, with token pricing remaining unchanged.
CognitionSep 28Cognition launched a beta for Devin Mobile, providing a mobile interface for its AI software engineer. The app runs in the cloud or on a local machine, tests in its own browser, and executes tasks until a pull request is ready to merge. Access is currently available through a waitlist at the official Devin website.
CognitionSep 25Cognition has reached $1 billion in annualized revenue run rate. The company attributes this milestone to its customers, including engineering teams at GE Aerospace, Rivian, Rohlik, and Exa, who use Devin to manage software development lifecycles. Devin became generally available less than two years ago, marking a shift toward enterprise-scale adoption of autonomous software engineering agents.
CognitionSep 23Cognition launched native Microsoft Teams support for Devin, enabling direct messaging, group chats, and channel-triggered automations. A new first-party Microsoft 365 integration connects Devin to Outlook, Calendar, OneDrive, SharePoint, and directory services. These integrations use per-user, delegated permissions with gated consent, ensuring Devin accesses only the data authorized by the signed-in user.
CognitionSep 22Cognition has integrated OpenAI's GPT-6 Sol and GPT-6 Luna into Devin Desktop and Devin CLI. On Cognition's FrontierCode 1.1 benchmark, Sol matches GPT-5.6 Sol's score at 61% lower cost per task. Luna scores higher than its predecessor at a quarter of the cost, becoming the cheapest model on the leaderboard at under $0.10 per task.
CognitionSep 22Cognition has integrated Anthropic's Claude Opus 5.5 into Devin Desktop and Devin CLI. The model now ranks first on Cognition's FrontierCode 1.1 Extended benchmark with a 65.3% score, displacing Fable 5. It performs tasks with less exploration and higher verification discipline while costing less than one-tenth of Fable 5 per rollout.
CognitionSep 22Cognition added xAI's Grok 4.7 to Devin Desktop and the Devin CLI. On the FrontierCode 1.1 benchmark, the model scores 59.4% on Extended, showing strength in hard backend engineering tasks. It trails Grok 4.6 on the aggregate due to a tendency to over-scope, generating larger diffs with unrequested changes.
CognitionSep 21Cognition launched Devin Cloud in Terminal: devin --cloud now creates, steers, and resumes cloud sessions directly from the CLI. The update also introduces devin ssh, providing direct access to Devin's dedicated virtual machine for code editing, port forwarding, and file management. SWE-2 sessions are free in Devin Cloud until October 8.
CognitionSep 17Cognition gave its Devin AI software engineer a Ramp card and a single goal: make money. The agent autonomously ran cold outreach, built payment portals, and experimented on a business plan. Through these self-directed actions, Devin successfully generated $75, demonstrating its ability to operate as an economic actor beyond its standard software engineering tasks.
CognitionSep 16Cognition launched Code Scans for Devin, enabling codebase-wide audits for engineering goals. Powered by Agentic MapReduce, the feature autonomously investigates code, reports findings, and opens pull requests. In testing, Code Scans reduced Dioxus compilation time by 64%. The feature is available by typing /scan in the Devin webapp.
CognitionSep 15Cognition launched macOS support for Devin, providing an autonomous Mac virtual machine for building and testing iOS and macOS applications. The agent uses the iOS Simulator to verify app behavior, sends screen recordings via Slack, and generates TestFlight links for deployment. This environment enables Devin to autonomously reproduce issues, iterate on code, and verify fixes in a native environment.
CognitionSep 15Cognition and AWS entered a multi-year Strategic Collaboration Agreement to help enterprises deploy autonomous engineers in production. Devin is now available on AWS Marketplace, supporting modernization of legacy systems and backlogs within dedicated VPC environments. Mercedes-Benz used the platform to reduce an eight-month COBOL modernization project to eight days, while another automaker cut costs by 73%.
CognitionSep 11Cognition introduced Fusion, a multi-model harness for Devin CLI and Desktop that pairs a frontier lead model for planning with a cost-effective sidekick for execution. This architecture maintains frontier performance while reducing costs by up to 39% across coding benchmarks. The lead model retains control, reviewing work and managing the session to ensure quality and efficiency.
CognitionSep 10Cognition launched Devin Voice, a conversational interface for its AI software engineer. Powered by GPT-Live and the SWE-2 model, the feature supports verbal interaction to explore ideas, test approaches, and hand off tasks. The interface includes microphone controls, audio silencing, and background navigation, with all conversation history saved directly to the active session.
CognitionSep 10Cognition launched SWE-2, its most advanced coding model, achieving 50.0% on FrontierCode 1.1 Main. By scaling reinforcement learning to multi-trillion parameters, SWE-2 matches Fable 5.1 performance at 64% lower cost. The model introduces support for configurable effort levels and is available today in Devin Desktop and CLI, with a one-month free period for Pro, Max, and Teams subscribers.
CognitionSep 9Cognition factored RSA-260, a 260-digit number, using a new GPU-optimized lattice siever built by its AI software engineer Devin. This implementation makes factoring 10x cheaper than the previous state of the art, costing approximately $400,000 in GPU time. The project demonstrates Devin’s ability to autonomously handle complex performance engineering and cluster management for large-scale computational research.
CognitionSep 8Cognition raised over $2B in Series E funding at a $48B valuation, led by Andreessen Horowitz and Accel. The company reports run-rate revenue growth from $492M in May to nearly $900M. Devin now features autonomous alert remediation via Slack and a Security Swarm for vulnerability triage, supporting enterprise engineering teams at NVIDIA, GE Aerospace, Citi, and Mercedes-Benz.
CognitionSep 3Cognition is adding OpenAI's GPT-6 Astra to Devin. On the FrontierCode 1.1 benchmark, Astra performs within 0.4 points of Fable 5 at a 64% lower cost. The model also sets a new SOTA on internal testing benchmarks, producing more comprehensive reports. Enterprise customers in OpenAI's Daybreak Program have access today, with broader rollout following in the coming days.
CognitionSep 1Devin, Cognition's AI software engineer, now runs on Fable 5.1, dropping task costs 54% thanks to Anthropic's 4x reduction in prompt-cache pricing (to $0.25/M cached tokens). Since 95%+ of Devin's tokens are cached, the pricing change has outsized impact. Cognition's Fusion multi-model harness hits the same FrontierCode 1.1 score for $1.43 per task — 47% cheaper than Fable 5.1 solo.
Devin DesktopAug 31Cognition updated the Devin CLI with a unified model picker. The interface integrates model selection, reasoning effort adjustment, and cost comparison into a single view. Accessible via the /model command, the update provides direct control over model behavior and efficiency for coding tasks.
CognitionAug 26Cognition overhauled the Devin webapp with a rebuilt chat renderer that reduces loading lag by 80% and speeds up long session openings by 55%. The update adds a customizable sidebar with compact views and grouping, a new ⌘K command menu for keyboard navigation, and native management for nested subagents. These changes improve performance and navigation for large-scale coding sessions.
CognitionAug 22Cognition is applying a 20% price reduction from OpenAI to its existing 70% discount for GPT-5.6 Sol, resulting in a 76% total discount off list prices. This pricing, effective through October 3, 2026, brings the model to $1.20 per 1M input tokens and $6 per 1M output tokens for users on Devin Desktop and Devin CLI.
CognitionAug 22Cognition reports that GE Aerospace is using Devin to build aviation software. One engineering team nearly doubled its output while developing systems for route optimization, fuel reduction, and anomaly monitoring. The deployment supports the company's goal to increase flight capacity without adding new aircraft, runways, or pilots.
CognitionAug 22Cognition introduced Slack Code in Devin, where the agent proactively creates dedicated code channels for complex tasks. When Devin identifies a task requiring more space than a thread, it automatically ports the session to a new channel. These channels feature a status chip, context bar, and artifacts tab to keep team work focused and transparent.
CognitionAug 13Cognition integrated Gemini 3.7 Flash into Devin Desktop and CLI. On the FrontierCode 1.1 benchmark, the model reaches Claude Sonnet 5-level performance at less than half the cost. It excels at tightly scoped refactors and matching repository conventions. A 50% discount on the model is available within Devin through August 27, 2026.
CognitionAug 13Cognition integrated MongoDB Atlas into Devin, enabling the AI agent to autonomously query, manage, and provision database clusters during software engineering tasks. This integration removes the need for manual database setup or static schema context, allowing the agent to interact directly with live data environments while building and testing applications.
CognitionAug 12Cognition integrated xAI's Grok 4.6 into Devin Desktop and the Devin CLI. On the FrontierCode 1.1 Extended benchmark, the model achieves a 61.3 score, surpassing GPT-5.6 Sol while trailing Claude Opus 5 and Claude Fable 5. The model demonstrates specific strengths in code exploration, root cause analysis, repo convention matching, and testing rigor.
CognitionAug 4Cognition updated Devin Fusion, achieving a 4% increase in intelligence and a 27% reduction in costs on the FrontierCode 1.1 benchmark. These gains result from refinements to both the model harness and the underlying models. Devin Fusion is available now in Devin Cloud for engineering teams.
CognitionAug 1Cognition now enables Devin to build and test native iOS apps using Devin Outposts. By running sessions on macOS environments powered by Namespace Devboxes, Devin gains access to Xcode, the iOS simulator, and Apple M5 chips. The agent autonomously builds, tests, and verifies iOS applications, sending screen recordings and test reports directly for review.
CognitionJul 30Cognition updated Devin to natively support GitHub Stacked PRs. The agent automatically decomposes large tasks into smaller, independently reviewable diffs, addresses review comments across the stack, and handles rebasing for downstream changes. This update maintains stack coherence and CI verification, allowing for focused reviews of complex code changes without manual bookkeeping.
Frequently asked questions