GPT-5.5 Now Available on GitHub Copilot with 7.5x Premium Multiplier

OpenAI's GPT-5.5 is now rolling out on GitHub Copilot. According to the changelog, early testing shows strongest performance on complex, multi-step agentic coding tasks, resolving real-world coding challenges that previous GPT models couldn't handle.
Availability
- Plans: Copilot Pro+, Copilot Business, Copilot Enterprise
- IDEs: Visual Studio Code, Visual Studio, Copilot CLI, GitHub Copilot cloud agent, github.com, GitHub Mobile (iOS/Android), JetBrains, Xcode, Eclipse
- Rollout: Gradual — check back if not yet visible
Pricing
GPT-5.5 launches with a 7.5× premium request multiplier as promotional pricing. This means each request counts as 7.5 against your included monthly requests.
Enabling Access
For Copilot Enterprise and Business plans, administrators must enable the GPT-5.5 policy in Copilot settings. Pro+ users can select it from the model picker directly.
How to Use
- Open the model picker in your supported IDE or tool.
- Select
GPT-5.5from the list. - Use as you would any other Copilot model — but expect better handling of multi-step agentic tasks.
Expect improved performance on complex debugging, multi-file refactoring, and long-context agentic workflows.
📖 Read the full source: HN LLM Tools
👀 See Also

OpenClaw Docker users: 2026.3.13 update missing Docker tags
OpenClaw version 2026.3.13 has been released, but Docker users should avoid updating as the Docker image lacks both 'latest' and '2026.3.13' tags. Users running from npm or git are unaffected.

Why One Developer Leaves AI Co-Author Tags in Commits
A developer explains why they deliberately include 'Co-Authored-by: Claude' in their Git commits, comparing it to EXIF data in photography and discussing the challenges of surgical AI-assisted code changes.

Amazon Employees Mock Company's AI on Slack, Call It 'Sloppenheimer'
Amazon employees have a Slack channel for memes mocking the company's faulty AI coding product, calling its output 'slop' and joking about failed AI motivation efforts.

AIME 2026 Results: Both Open and Closed Models Score Above 90%
AI models achieve remarkable 90%+ scores on AIME 2026, with DeepSeek V3.2 running the entire test for just bash.09.