JP — ITmedia

GitHub Releases "Copilot Fusion" Generally Available, Selecting the Best AI Model for Each Task to Balance Performance and Cost

AINovember 4, 20261 min read

GitHub has officially released Copilot Fusion, a tool that picks one or more AI models for each coding task. The goal is strong results at a lower cost.

What happened

At its GitHub Universe 2026 event in San Francisco (October 28-29), GitHub announced the full release of GitHub Copilot Fusion. It is a multi-model feature. It looks at each task and decides whether to use one AI model or to combine several. This takes the job of choosing models away from developers and helps them balance performance and cost.

Key facts

Copilot Fusion has 3 patterns: Single, Cascade, and Critique.
Single uses one chosen model to solve the task.
Cascade: a cheaper model makes a draft, a checker model reviews it, and a stronger model is called only if needed.
Critique: one model writes a draft, a model from a different family reviews it, and the first model revises it one time.
Four levels set the balance of quality and cost: Fusion Standard (everyday work), Fusion High (large coding tasks like complex bug fixes), Fusion Max (the deepest reasoning), and Fusion Fast (users cannot choose it).

Some background

Many AI coding tools let developers pick a model by hand. Different models are good at different things, and strong models cost more to run. Copilot Fusion tries to make that choice automatically, so each task gets a model that fits its difficulty.

Why it matters

Developers often do not know which AI model is best for each job. An automatic choice can save money and time without losing quality. As AI tools spread, cost control and quality will matter more for companies that build software.

What to watch

In public preview, Fusion Standard can run tasks on locally installed MAI Code 1.1 Flash or Qwen3.8 Flash Next models.
Whether real users see better results at lower cost.
How well the automatic model choice works on large, multi-file coding tasks.

Glossary

GitHub Copilot Fusion: A GitHub feature that automatically picks and combines AI models for each coding task.
GitHub: A company whose website hosts software code and offers developer tools, including Copilot.
Cascade: A pattern where a cheap model tries first, and a stronger model is used only if the check says it is needed.
Critique: A pattern where one AI model writes a draft and a different AI model reviews it, then the draft is fixed once.

ByteDance reportedly tightens access to Claude and Codex, even for employees paying out of pocket

According to recent reports, ByteDance has since early this month restricted employees from using personally purchased accounts for third-party AI tools such as Claude and Codex to access internal resources, covering all L4-level code repositories and Feishu documents. ByteDance vice president of technology Hong Dingkun said the move is driven mainly by data security concerns, and denied that it aims to force employees onto the company's own products. The restrictions are being rolled out in phases and are expected to affect thousands of technical employees.

AIITmediaNov 4

Google to Turn Off AI Features for School Students by Default; Critics Say Google Search Without AI Can't Be Chosen

Google announced on November 3 (local time) that AI features, including Gemini and Google Search, will be off by default for K-12 student accounts until school administrators explicitly enable them. However, AI features built into Google Search cannot be disabled on their own, so schools that opt out of AI would also lose access to Google Search, drawing criticism. According to the New York Times, the change is expected to take effect in early 2027.

AIThe VergeNov 4

Please stop asking me when the AI bubble will pop

The Verge's Elizabeth Lopatto says she doesn't know when the AI bubble will burst, and points to a Bloomberg roundup of warnings from the 2026 Global Financial Leaders' Investment Summit in Hong Kong. PIMCO's CEO said an AI downturn is coming and "then things are going to break," while Apollo also flagged AI risk. Lopatto notes that heavy AI-related bond sales from Meta, Oracle, Alphabet and Microsoft continue, even as caution grows.

AITechCrunchNov 4

Anthropic is working on a system to prove AI models’ identity

Anthropic is developing FlashProve, an open source system that uses cryptographic proofs to verify which AI model produced a given output. Labs would publish a model "fingerprint" that does not reveal model weights, allowing users and third parties to check claims about which model was used, including with commercial models. The research is ongoing, and Anthropic has no plans to deploy the system at Anthropic or any other lab; its code and specification are publicly available on GitHub.