# GitHub Universe 2026: Copilot Fusion, the Copilot App and What Businesses Gain By [Eoin Villarin](https://amvprosox.com/about). Published 2026-10-29. Canonical: https://amvprosox.com/blog/github-universe-2026-copilot-fusion > GitHub Universe 2026 launched Copilot Fusion, Copilot app Projects and Copilot inside GitHub Enterprise. Here is what changed and what US businesses should ask their developers. ## What was announced at GitHub Universe 2026? GitHub Universe 2026, held October 28 to 29 in San Francisco, launched Copilot Fusion (generally available on all Copilot plans), Projects and computer use previews in the Copilot app, Copilot code review Max, AI Scan for repositories, and an option for GitHub Enterprise customers to use Copilot without separate seats, paying only for AI usage. **Key takeaways** - GitHub Universe 2026 ran October 28 to 29, 2026, at Fort Mason Center in San Francisco. - Copilot Fusion became generally available on October 28, 2026, on all Copilot plans, with a limited-time 30% model cost discount for paid plans. - GitHub says Fusion Max scored 60 on the Artificial Analysis Coding Agent Index at about half the API cost per task of Claude Code with Opus 5.5 Max. - GitHub Enterprise Cloud customers working with an account team can now get Copilot included and pay only for AI usage, while Copilot Standalone costs $10 per user per month. - Octoverse 2026 reports that more than 40% of pull requests on GitHub started with an agent's first draft by September 2026, up from 17% in March. GitHub Universe 2026 opened in San Francisco on October 28 with a clear message: AI agents now do a large share of the coding on GitHub, and GitHub wants Copilot to manage that work from issue to merge. The headline launches were Copilot Fusion, which picks AI models for you, a bigger Copilot app with multi-repository Projects and computer use, and a new deal that lets GitHub Enterprise customers use Copilot without buying separate seats. This guide is for US business owners, marketing leads and founders who pay for websites, stores or software but do not live inside GitHub. You get what was announced, what it costs, what is still in preview, and what to ask your developer or agency to change. Facts below come from [GitHub's Universe 2026 keynote post](https://github.blog/news-insights/company-news/github-universe-2026-evolving-every-layer-of-the-stack/), the official [GitHub changelog](https://github.blog/changelog/) and the [Octoverse 2026 report](https://github.blog/news-insights/octoverse/octoverse-commits-pushes-reach-new-highs-in-2026-as-developers-build-with-agents/). Benchmark and usage numbers are GitHub's own figures, and I label them that way. ## GitHub Universe 2026 at a Glance The event ran October 28 to 29 at Fort Mason Center, according to [GitHub's schedule announcement](https://github.blog/news-insights/company-news/your-guide-to-github-universe-2026-is-here-the-schedule-just-launched/). GitHub published about 30 changelog entries on day one. These matter most for teams that build and run websites. | Announcement | Status on Oct 28, 2026 | Who can use it | Why it matters | | --- | --- | --- | --- | | Copilot Fusion | Generally available | All Copilot plans | Routes each task to one or more AI models automatically | | Copilot app: Projects | Public preview | Copilot app users | Runs agents across several repositories toward one goal | | Copilot app: computer use | Public preview | macOS and Windows | Lets Copilot operate desktop apps that have no API | | Copilot included with GitHub Enterprise | Available via sales teams | GitHub Enterprise Cloud | No separate Copilot seats, pay for AI usage instead | | Copilot code review Max | Public preview | Pro, Pro+, Max, Business, Enterprise | Several review agents check high-risk changes | | AI Scan for repositories | Public preview | Public repos, plus private repos with GitHub security products | Agentic security scan of the whole codebase | | Actions workflow lockfiles | Generally available | Free, Team, Enterprise Cloud | Pins the exact code your build pipeline runs | ## What Is GitHub Copilot Fusion? Copilot Fusion is a new option in the Copilot model picker. Instead of choosing a model yourself for every task, you pick Fusion and an effort level, and Copilot decides which model or combination of models should handle each prompt. According to the [Copilot Fusion changelog](https://github.blog/changelog/2026-10-28-route-and-orchestrate-models-with-github-copilot-fusion/), it merges GitHub's earlier HydraFusion research preview with Copilot's auto model selection, and it is generally available on all Copilot plans. ![GitHub Universe 2026 Copilot Fusion menu with Standard, High and Max effort levels](https://amvprosox.com/api/img/1793276914703-vv549k) *Copilot Fusion appears in the model picker with three effort levels. Image: GitHub* ### The three effort levels - **Standard** favors versatile models for everyday coding. - **High** favors more capable models for harder tasks and deeper reasoning. - **Max** favors the most capable models for the most complex work. ### How Fusion combines models GitHub says Fusion can use three patterns. In a **single** run, one model does the whole job. In a **cascade**, a cheaper model drafts an answer and a quality check either accepts it or escalates to a stronger model. In a **critique**, one model drafts, a second reviews, and the first revises. The point is to stop paying top-tier prices for simple work. ![Diagram of the Single, Cascade and Critique execution patterns in GitHub Copilot Fusion](https://amvprosox.com/api/img/1793276915240-2atulp) *GitHub's diagram of the three ways Fusion can split a task between models. Image: GitHub* ### Price and performance claims Fusion is billed per model it uses, at each model's standard rate, and paid Copilot plans get a 30% discount on model costs with Fusion for a limited time. GitHub reports that Fusion Max scored 60 on the Artificial Analysis Coding Agent Index, within two points of Claude Code running Opus 5.5 Max, while using about half the average API cost and tokens per task and about 40% less time. These are vendor figures, so treat them as a starting point for your own testing. Fusion works in VS Code, the GitHub Copilot app, Copilot CLI and JetBrains IDEs. Admins stay in control: Fusion only uses models the organization has enabled, and a model can be removed from Fusion without turning it off for everyone. A separate [experimental preview](https://github.blog/changelog/2026-10-28-github-copilot-fusion-now-supports-selection-of-local-models/) lets Fusion route between on-device and cloud models. ## What Is New in the GitHub Copilot App? The Copilot app, which entered technical preview in May, is GitHub's desktop home for agent work. GitHub describes it as built for taking work from issue to merge in one place. The [Copilot app Universe release notes](https://github.blog/changelog/2026-10-28-github-copilot-app-universe-releases/) list three big additions. - **Projects (public preview):** you describe a goal, attach context such as repositories, design mocks, Slack threads or meeting notes, and Copilot runs as many agents as the work needs, returning pull requests for review. Project automations can handle repeat jobs. - **Computer use (public preview on macOS and Windows):** Copilot can read and click through desktop apps that have no API, CLI or MCP integration. It asks for approval before controlling an app, and organizations can switch the feature off. - **Work around your code:** Copilot can pull context from tools like Slack and Figma to draft specs, and it can now [generate and edit images](https://github.blog/changelog/2026-10-28-generate-and-edit-images-directly-in-copilot/) from a text prompt. GitHub also announced **session continuity**, so one cloud agent session can move between a phone, VS Code and github.com, and **Copilot Automations**, cloud workflows that run on a trigger or schedule for jobs like issue triage, which GitHub says can now propose improvements for a person to approve. ## How Much Does GitHub Copilot Cost After Universe 2026? Pricing changed most for larger companies. According to the [GitHub Enterprise changelog](https://github.blog/changelog/2026-10-28-github-enterprise-now-includes-access-to-github-copilot), enterprises that work with a GitHub or Microsoft account team can now get Copilot included with GitHub Enterprise, so every licensed user has access and there are no separate Copilot seats to buy. Instead, AI usage is metered from the first request against an enterprise budget. | Option | What you pay | Notes | | --- | --- | --- | | Copilot included with GitHub Enterprise | AI usage against your budget, no seat fee listed | Through account teams now, self-serve coming in stages | | Copilot Standalone | $10 per user per month | For enterprise members without a GitHub Enterprise license | | Copilot Business and Enterprise | Existing plans | Still sold, current self-serve subscriptions unaffected | If the AI budget runs out, code completion and next edit suggestions keep working, but Copilot Chat, code review and cloud agents stop until budget is added. AI spend now behaves more like a utility bill than a seat price, so budget alerts matter. ## Security and Code Quality Updates From GitHub Universe 2026 The quieter announcements may matter more for the safety of your website than the agent features. - **Copilot code review Max (public preview)** runs several independent review agents, each on a model from a different provider, then reconciles their findings into one review. GitHub positions it for risky changes such as authentication logic, per the [code review changelog](https://github.blog/changelog/2026-10-28-copilot-code-review-merge-with-confidence-with-max-and-approvals). - **Copilot approvals (generally available)** let Copilot approve a pull request in a way that counts toward required approvals. It is off by default, new commits cancel the approval, and admins can limit which file paths Copilot may approve. - **AI Scan for repositories (public preview)** scans the whole repository, including languages CodeQL does not cover, and does not need a Copilot license, according to the [AI Scan changelog](https://github.blog/changelog/2026-10-28-ai-scan-for-repositories-in-public-preview). - **Workflow lockfiles (generally available)** record the exact commits your GitHub Actions run, so a moved tag cannot silently change your build. GitHub notes in the [lockfiles changelog](https://github.blog/changelog/2026-10-28-github-actions-workflow-lockfiles-are-generally-available) that a lockfile does not prove a dependency is safe. - **GitHub Agentic Workflows (generally available)** turn plain-language Markdown instructions into standard Actions workflows that run coding agents for tasks like triage and CI failure analysis. ## What Octoverse 2026 Says About AI and Software Work GitHub released its annual Octoverse report the same day. Its numbers explain why Universe was all about agents. GitHub reports that by September 2026 more than 40% of pull requests on GitHub started with an agent's first draft, up from 17% in March, and that 110 million agent-started pull requests were opened in September alone. ![Octoverse 2026 statistics showing over 40% of pull requests are agent generated and 110 million in September 2026](https://amvprosox.com/api/img/1793276915818-d2hl7l) *Octoverse 2026 tracks the rise of agent-started pull requests. Image: GitHub* GitHub also reports more than 225 million developers, with the United States still the largest community at 38.3 million, and says 61% of Copilot users who chose a model themselves used more than one in September. ## What GitHub Universe 2026 Means for Your Business If you run a US small business, an ecommerce brand or an agency, you will probably never open the Copilot app yourself. But the people who build your website will, and these changes affect the speed, cost and safety of the work you pay for. Here is what I would do this quarter. - **Ask your developer how they use AI agents.** Agents now draft a large share of code. That is fine, but you should know which tools touch your code, which models see it, and who reviews the result before it goes live. - **Keep a human approval on risky changes.** Copilot can now approve pull requests. For checkout, login, payment and customer data code, require a named person to approve as well. Copilot approvals can be limited by file path, so this is easy to set up. - **Turn on the free and cheap security wins.** Push protection, Dependabot and workflow lockfiles cost little and stop real problems, such as a leaked API key or a tampered build step. If your store or web app is on GitHub, ask for these to be enabled. - **Put your website code under version control.** Many small business sites, especially custom WordPress themes and Shopify themes, still live only on a server. Moving the code into a private GitHub repository gives you history, backups and a safe way to use AI review tools. - **Treat AI usage as a budget line.** Fusion and the new Enterprise model bill by usage. Set spending limits and alerts so a busy month does not surprise you. - **Expect faster delivery.** If agents cut routine work, small fixes should get quicker and cheaper for you too. ### Where AI agents still need an expert Agents are good at well-defined tasks with clear tests. They are weaker at judgment calls: choosing an architecture, spotting a business rule that is wrong, or knowing which third-party script is slowing your checkout. I covered this balance in [vibe coding vs hiring a web developer](https://amvprosox.com/blog/vibe-coding-vs-hiring-web-developer), and I explain my own review process in [how I use AI coding agents to ship client websites faster](https://amvprosox.com/blog/ai-coding-agents-freelance-developer-workflow). If your site runs on WordPress, the plugin risks in [WordPress security in 2026](https://amvprosox.com/blog/wordpress-security-plugin-vulnerabilities-2026) are a good reason to pair AI scanning with human review. ## Frequently asked questions ### When was GitHub Universe 2026? GitHub Universe 2026 took place on October 28 and 29, 2026, at Fort Mason Center in San Francisco. GitHub published its main announcements on October 28, including Copilot Fusion, Copilot app Projects, computer use, Copilot code review Max and the option to include Copilot with GitHub Enterprise. Most launches went live the same day, some as public previews. ### Is GitHub Copilot Fusion free? Copilot Fusion is available on all Copilot plans, but usage is not free. Each model Fusion uses is billed at that model's standard rate. For a limited time, users on paid Copilot plans get a 30% discount on model costs when they use Fusion, so it is worth testing now on real tasks to compare cost and quality. ### Do GitHub Enterprise customers still need to buy Copilot seats? Not necessarily. Since October 28, 2026, GitHub Enterprise Cloud customers working with a GitHub or Microsoft account team can get Copilot included with every licensed user and pay only for AI usage. Self-serve customers cannot switch yet. Copilot Business and Copilot Enterprise plans remain on sale, and existing subscriptions are not affected. ### Can Copilot now approve and merge pull requests by itself? Copilot can now approve pull requests, and that approval can count toward required reviews, but the feature is off by default. An admin must enable it, new commits cancel the approval, and admins can restrict which file paths Copilot may approve. For payment, login and data code, keeping a human reviewer is still the safe choice. ### What does GitHub Universe 2026 mean for small businesses? It means the developers who build your website will increasingly use AI agents for routine work. That can lower costs and speed up fixes, but it also raises the importance of code review, security scanning and budget controls. Ask your developer how they use these tools and make sure a person still approves important changes. ## Make AI Work for Your Website, Not Against It GitHub Universe 2026 confirmed that AI agents are now a normal part of building software. The winners will be businesses whose developers use these tools to move faster while keeping review, security and costs under control. The tools are ready. The process around them is what makes the difference. If you want a developer who uses AI agents responsibly to build, fix or maintain your Shopify store, WordPress site or custom web app, [tell me about your project](https://amvprosox.com/#contact-form). I work with US businesses across time zones, and you can see the results in my [case studies](https://amvprosox.com/case-studies). I will reply within 24 hours.