GitHub ships Copilot Vision and in-editor browser tools to general availability

466 reads

Read the original article →

What happened

GitHub announced on July 1 that Copilot Vision is now generally available, letting users attach images and PDFs to chat prompts so Copilot can reason about them alongside code. It is available across all Copilot tiers, from Free to Enterprise, and is on by default. GitHub said image and PDF attachments are retained for roughly 24 hours on Business and Enterprise plans to provide the service.

The same day, GitHub moved browser tools for Copilot in VS Code from preview to general availability, extending Copilot's agentic workflows in the editor. Both shipped as part of a single day of Copilot releases.

Why it matters

Multimodal input brings design mockups, screenshots and PDFs directly into the coding loop, which helps when reproducing a UI from an image or debugging from a screenshot. In-editor browser tools push Copilot further toward acting inside the editor rather than only suggesting.

MintedBrain take

The practical value of Vision depends on how well it grounds answers in the attached image rather than guessing. For teams on Business or Enterprise plans, note the roughly 24-hour retention of attachments and check it against your data-handling policies before pasting anything sensitive.

Related articles

  • UN opens its first Global Dialogue on AI Governance in Geneva

    The United Nations convened its first Global Dialogue on AI Governance in Geneva on July 6, a two-day session established by the UN General Assembly as the first intergovernmental platform dedicated to AI. The UN said it brings together all 193 member states alongside private-sector and civil-society participants. The UN's Independent International Scientific Panel on AI presented a preliminary report to governments.

  • UN science panel warns AI is outpacing safeguards as governance summit nears

    In a July 5 feature previewing its Geneva meetings, UN News published interviews with the co-chairs of the new Global Dialogue on AI Governance and the UN's Independent International Scientific Panel on AI. Panel co-chair Yoshua Bengio said AI capabilities are outpacing scientific understanding and that science currently cannot guarantee advanced AI will not cause catastrophic harm. Co-chair Maria Ressa described AI-amplified disinformation as an 'information Armageddon.'

  • xAI makes Grok Speech-to-Text and Text-to-Speech APIs generally available

    xAI moved its Grok Speech-to-Text and Text-to-Speech APIs to general availability, giving developers audio transcription across 25 languages with batch and streaming modes plus natural-sounding speech generation. The move targets enterprise voice-agent developers building on the Grok platform. It is part of xAI's broader July 2026 developer-API expansion.

  • Anthropic moves to close loopholes Chinese firms use to access Claude

    The Financial Times reported Anthropic has stepped up efforts to detect and shut down unauthorized Claude access by Chinese companies, identifying workarounds such as routing employee accounts through overseas subsidiaries and reimbursing engineers for personal subscriptions accessed via VPNs. Anthropic's detection now monitors indicators like user time zones and targets relay services. The company frames the activity as distillation attacks meant to extract Claude's capabilities.

References

This article was originally published at GitHub. For the full piece, read the original article.

Discussion

  • Loading…

← Back to News