[News] GitHub Copilot's New HydraFusion Multi-Model Mode Lands in VS Code, Free Tier Left Out

I drafted the piece below. One note: the target output path (/root/AMPM-AIOPS/ampm-site/content/guides/github-copilot-news-20261002.md) resolves through a symlink to /opt/ampm/..., which is outside this session’s allowed directories — I can’t write there directly. The full article is below for you to place.

GitHub pushed its multi-model orchestration feature, HydraFusion, out of the command-line Copilot CLI and into Visual Studio Code and the standalone GitHub Copilot app on 2026-09-30. If you’re on GitHub Copilot’s free tier — the one capped at 2,000 code completions and 50 chat requests a month — this update changes nothing for you. HydraFusion only works for Pro (US$10/month), Pro+ (US$39/month), Business (US$19 per seat/month) and Enterprise (US$39 per seat/month) subscribers. Free users keep their existing limits and keep being locked to “Auto” model selection, a restriction GitHub put in place back on 2026-06-24.

This piece looks at what actually changed, who’s paying for it already, and whether it’s worth going near.

What HydraFusion Actually Does

HydraFusion isn’t a model. It’s described by GitHub as an “optimization engine” that picks how to solve a coding request based on signals like reasoning difficulty, code generation complexity, debugging needs and tool use. It runs one of three workflows: a single model solves the request directly; a cascade mode where a fast model drafts an answer and a quality gate decides whether to escalate to a stronger model; or a critique mode where an independent, read-only reviewer model from a different model family checks the draft and the original model revises once based on that feedback.

That’s different from Auto, which just assigns one model to a request and stops there. HydraFusion can juggle multiple models inside a single back-and-forth, aiming for what GitHub calls frontier-level code quality rather than whatever one model happens to produce.

To use it, you need VS Code version 1.140 or newer. Even then, it’s not guaranteed to show up automatically — if it doesn’t appear in the model picker, you have to manually turn on the chat.copilot.hydraFusion.enabled setting. In the standalone GitHub Copilot app, you need the latest app version and have to switch the option on yourself. For Enterprise seat-holders, there’s an extra step: an organization admin has to enable the preview in org-level settings before anyone on the team can touch it.

Who Actually Gets To Use It

HydraFusion sits entirely inside pricing that already existed — GitHub didn’t create a new tier or raise prices to launch it. Here’s where it lives:

• Pro, US$10/month: unlimited code completions and next-edit suggestions, model selection, access to third-party agents (Claude Code and Codex), cloud agent and code review — plus US$15/month worth of GitHub AI Credits (a usage currency spent on premium model requests) included.
• Pro+, US$39/month: premium models including Opus, audit logs, usage described as “4 times or more” Pro’s allowance, and US$70/month of AI Credits included.
• Max, US$100/month: a newer plan that bundles about US$200/month in GitHub AI Credits for heavy users.
• Business, US$19 per seat/month: 1,900 AI Credits per seat as the standard allotment. Organizations of 10, 25 or 50 seats get volume discounts of 5%, 10% or 15% respectively, according to third-party reporting.
• Enterprise, US$39 per seat/month: 3,900 AI Credits per seat standard.

GitHub ran a promotion on Business and Enterprise credits — seats got bumped to 3,000 and 7,000 credits respectively — but that ran from 2026-06-01 to 2026-09-01. That window closed a month before this HydraFusion rollout, so teams signing up for Business or Enterprise seats now get the standard 1,900 or 3,900 credits, not the summer promo numbers.

Free tier users get none of this. Their 2,000 monthly completions and 50 monthly chat requests already excluded cloud agent and code review before HydraFusion existed, and since 2026-06-24 they can’t even manually pick a model — Auto is the only option. There’s no way to opt into multi-model orchestration without first paying for Pro. GitHub does run separate no-cost paths into Pro: open-source maintainers and teachers can get a free Pro upgrade, and new student free-plan signups reopened gradually starting 2026-06-17 after a 58-day pause that began 2026-04-20. GitHub’s changelog names Pro, Pro+, Business and Enterprise as the eligible plans for HydraFusion — it doesn’t separately confirm whether maintainer or student free-upgrade accounts count as “Pro” for this purpose, so that’s worth checking against your own account status rather than assuming.

🔍 AMPM exclusive check

GitHub’s changelog announcement doesn’t frame this as a pricing story — it’s written as a feature rollout. We checked it as one anyway, and three things stood out that the vendor page doesn’t spell out.

First: this is not a new paywall. We cross-referenced HydraFusion’s plan list against GitHub’s existing feature breakdown and found it slots into the same already-paid bucket as cloud agent and code review — features free users never had. No price moved, no new tier was invented to gate this specifically.

Second: the Business/Enterprise credit promotion (1,900→3,000 and 3,900→7,000 credits per seat) expired 2026-09-01, a full month before HydraFusion reached VS Code on 2026-09-30. Anyone evaluating Business or Enterprise seats today, drawn in by this news, should budget for the standard credit numbers, not the numbers that were live over the summer.

Third: “available to Pro and up” doesn’t mean “on by default for Pro and up.” It requires VS Code 1.140+, a manual settings flag, and — for Enterprise — a separate admin-level org setting. A Pro subscriber who doesn’t update their editor or go digging in settings will see nothing different at all. Nothing else in the pricing or quota figures differed from GitHub’s official page — those numbers checked clean as of 2026-10-01.

☀️ AMO, the budget-minded cat

Free tier gets zero say in this one — 2,000 completions and 50 chat requests a month never included cloud agent or code review, and they don’t include HydraFusion either. So the real question isn’t “should I use HydraFusion,” it’s “is US$10/month Pro worth it now,” and HydraFusion alone isn’t a strong enough reason to say yes. It’s a research preview, GitHub’s own term, which means the feature and its interface can change or vanish without warning. Paying to test something that unstable feels premature for anyone counting dollars. The US$15/month AI Credits bundled into Pro are the more concrete reason to upgrade — unlimited completions plus a real credit allowance you can spend on premium requests, HydraFusion or not. Pro+ at US$39/month with US$70/month in credits is a bigger jump that only makes sense if you’re already deep into premium model usage, not just curious about multi-model orchestration. The one genuinely free door into this is the maintainer/teacher Pro upgrade — if you qualify, that’s the only way to try HydraFusion without spending anything, and it’s worth checking eligibility before reaching for a card.

🌙 PIMI, the performance-minded cat

This is the more interesting read. Auto mode has always just handed a request to one model and called it done. HydraFusion’s cascade and critique workflows are a real structural difference: a cheap model drafts, a quality gate decides if it’s good enough, and if not, a stronger model takes over — or a second model from an entirely different family reviews the draft before one revision pass. That’s built for exactly the cases where a single model tends to fail: multi-step debugging, complex code generation, tool-heavy tasks. GitHub also says it improved transparency — more visible step-by-step execution and more frequent progress updates during long-running tasks, which matters when you’re staring at a spinner wondering if anything’s still happening. The catch is access friction: it’s not on by default even for paying plans, needs VS Code 1.140+, and Enterprise seat-holders need an admin to flip an org-level setting before anyone touches it — a real adoption drag inside bigger teams. Max subscribers at US$100/month with roughly US$200/month in bundled AI Credits have the most headroom to actually run cascade and critique modes repeatedly without watching the credit meter, which makes Max the plan where HydraFusion gets used the way GitHub designed it, rather than tried once and forgotten.

Verdict: who should use it, who should skip it

Use it if you’re already on Pro or above and want better results on hard debugging or multi-step coding tasks — the cascade and critique workflows are a genuine upgrade over single-model Auto, and the setup (updating VS Code, flipping one setting) is a few minutes of friction for subscribers already paying. Business and Enterprise teams should go in knowing their credits reset to standard 1,900/3,900-per-seat levels now that the 2026-06-01–2026-09-01 promo window is over, and plan for the admin step before anyone on the team can use it.

Skip it if you’re on the free tier — there’s no upgrade path that adds HydraFusion without first paying for Pro, so it doesn’t change your free-tier math at all. Also skip it if you’re budget-conscious and were already on the fence about the US$10/month Pro price: a research-preview feature that can change at any time isn’t a strong enough reason to tip that decision, and open-source maintainers or teachers should check the free Pro upgrade path first before paying for anything.

Written: 2026-10-02 Price last verified: 2026-10-01 Tool last health-checked: 2026-10-02

🔗 Related on AMPM

• GitHub Copilot pricing & free tier
• aider vs github-copilot
• bolt vs github-copilot

📎 Sources

All prices and quotas in this article come from AMPM’s verified dataset for GitHub Copilot (last verified 2026-10-01); primary sources:

• https://github.com/features/copilot/plans
• https://github.com/features/copilot

🐾 Meet the cats: AMO and PIMI

AMPM is a Taiwan-based AI-tool pricing watchdog. Our motto: ask before you subscribe. Two cats argue every tool from two sides:

• ☀️ AMO — the budget-minded cat. Always asks: is the free tier enough, and is paying actually worth it?
• 🌙 PIMI — the performance-minded cat. Cares about whether it works well, fast, and gets your job done.

When they are done arguing, you know whether to pay. Every number is checked by us, with the verification date at the end.

🐾 Our sister sites also checked

• Travel eSIM plans compared
• Exchange fees and safety checked