DEVOPSdigest

GitHub announced the next evolution of its Copilot-powered developer platform.

Anchored on giving developers more agency over the tools they use, GitHub revealed that they’re bringing developer choice to GitHub Copilot by making it multi-model, enabling developers to select from industry-leading models including Anthropic’s Claude 3.5 Sonnet, Google’s Gemini 1.5 Pro, and OpenAI’s GPT-4o, o1-preview and o1-mini.

GitHub also introduced GitHub Spark, an AI-native tool to build personal, customized, and fully functional web apps entirely in natural language.

GitHub also envisioned the AI-native developer experience with substantial updates to GitHub Copilot in VS Code, Copilot Workspace, GitHub Models, and Copilot Autofix, bringing AI functionality to the entire GitHub platform from issues, to pull requests, to builds.

GitHub Copilot

Developers using GitHub Copilot in Visual Studio Code and github.com can now choose from an array of industry-leading models including Anthropic’s Claude 3.5 Sonnet, Google’s Gemini 1.5 Pro, and OpenAI’s GPT-4o, o1-preview, and o1-mini. Developers can toggle between models during a conversation with Copilot Chat to choose the right model for the right use case, or continue to let Copilot use its powerful default. With this multi-model approach, GitHub is enabling developers to build with an array of leading models in the workflows they’re accustomed to.

Anthropic’s Claude 3.5 Sonnet will be available via GitHub Copilot starting today, and Google’s Gemini 1.5 Pro will be available in the coming weeks. GitHub will continue to enable developer choice in partnership with leading model providers, and will bring multi-model choice across many more of GitHub Copilot’s surface areas and functions soon.

"In 2024, we experienced a boom in high-quality large and small language models that each individually excel at different programming tasks. There is no one model to rule every scenario, and developers expect the agency to build with the models that work best for them," said GitHub CEO Thomas Dohmke. "It is clear the next phase of AI code generation will not only be defined by multi-model functionality, but by multi-model choice. Today, we deliver just that."

GitHub Spark

GitHub Spark makes it easy for developers of all skill ranges to bring ideas to life by using natural language to build micro apps called a “spark.” Sparks are fully functional micro apps that can integrate AI features and external data sources without requiring any management of cloud resources.

Utilizing a creativity feedback loop, users start with an initial prompt using both OpenAI and Anthropic models, see live previews of their app as it's built, easily see options for each of their requests, and automatically save versions of each iteration so they can compare versions as they go. Experienced developers can directly make changes to the underlying code while consumers or novice developers can iterate entirely in natural language—the choice is theirs. Once a user is happy with their spark, they can automatically run it on their desktop, tablet, or mobile device, ultimately getting immediate value from their own creation. They can also share their sparks with customized access control, as well as allow others to remix their spark and build upon their creations.

“For too long, there has been an unscalable barrier of entry separating a vast majority of the world’s population from building software. This can change with GitHub Spark, our new AI-native tool to build applications entirely in natural language,” said GitHub CEO Thomas Dohmke. “With Spark, we will enable over one billion personal computer and mobile phone users to build and share their own micro apps directly on GitHub—the creator network for the Age of AI."

GitHub will continue to iterate GitHub Spark to make the tool as intuitive as possible for both general consumers and developers of all skill ranges.

AI-Native Experiences

From advancements to GitHub Copilot in VS Code to the next iteration of Copilot Workspace and GitHub Models, GitHub unveiled its vision of an AI-native and agentic developer experience powered by developer choice and control.

- Multi-file edit for GitHub Copilot in VS Code: Users can easily use Copilot Chat in VS Code to make edits across multiple files at the same time. In this new editing mode, Copilot implements complex changes across a variety of files within a project based on natural language prompts.

- GitHub Copilot Extensions for all users: Copilot Extensions allow developers to ask questions of any integrated developer tool including from leading developer tools and services like Atlassian Rovo, Docker, Sentry, and Stack Overflow. Users will also be able to build their own private extensions that work with their internal developer tooling. Copilot Extensions will be generally available in early 2025.

- GitHub Copilot for Xcode: The code completion capabilities of Copilot are now available in public preview for Xcode, empowering developers building apps across all Apple platforms.

- Get Copilot-powered feedback on your code: With a new code review capability, Copilot offers fast feedback on code in 30 seconds, so users can start iterating towards "ready to merge" while waiting for a human reviewer. Users can request a review from Copilot in Visual Studio Code, or on GitHub.com when they create a pull request.

- Copilot tailored to your preferences: Users can now specify custom instructions to personalize Copilot Chat responses in VS Code and Visual Studio based on their preferred tools, organizational knowledge, and coding conventions. Plus, developers can leverage additional context from their repositories, pull requests, issues, discussions, and the web via Bing integration to have an AI-native experience across GitHub.

More updates across the GitHub platform include:

- The next iteration of Copilot Workspace: Working closely with developers to understand where Copilot Workspace can provide even more value, GitHub has since rolled out over 100 changes, including a build and repair agent, detecting necessary changes for completion, and running commands to repair errors after Copilot Workspace has generated a code implementation. Additional key updates include brainstorming mode, integrations with VS Code, iterative feedback loops, deeper AI assistance, and even greater context and personalization.

- Expanded features with GitHub Models in public preview: Since launching the interactive model playground, more than 70,000 developers have experimented with their AI model of choice—from OpenAI o1 and Meta Llama 3, to Microsoft Phi and Cohere Command R—directly from GitHub. Starting today, AI engineers can leverage new capabilities, including side-by-side model comparison, support for multi-modal models, the ability to save and share prompts and parameters, and new cookbooks and SDK support in GitHub Codespaces.

- Copilot Autofix and security campaigns to fix vulnerabilities at scale: Following the general availability of Copilot Autofix, developers are already fixing code vulnerabilities more than three times faster. Copilot Autofix now includes security campaigns to help developers and security teams remediate vulnerabilities at scale, with the ability to triage up to 1,000 alerts at a time, as well as filter alerts by type, severity, repository, and team. Copilot Autofix also now integrates with partner tools, including ESLint, JFrog SAST, and Black Duck’s Polaris™ platform powered by Coverity®, so developers can streamline security workflows with their code scanning tooling of choice. Security campaigns and Copilot Autofix for partner tools are available today in public preview.