Big Tech AI Subscriptions: What Everyone Actually Charges

Scope: companies that train their own models and sell a self-serve consumer subscription. The companion pieces cover independent alternatives and Chinese services.

Prices as listed in early October 2026, USD/month unless noted. This tier churns: Microsoft retired Copilot Pro on 1 August 2026, Google cut its top tier by more than half, and Meta only started charging mid-2026. Verify before committing.

Prices

Two things stand out. Google has the cheapest serious entry at $4.99, less than a third of the $20 that OpenAI, Anthropic, Google and Microsoft all converge on for their main tier. And xAI is the most expensive to escalate — $30 for standard where others charge $20, up to $300.

For software development

Anthropic and OpenAI are the two serious choices, and the split is roughly: Claude for sustained work inside a real codebase (Claude Code ships with Pro, and terminal agents are where this work has moved), ChatGPT for breadth and the widest tooling ecosystem. At $20 they cost the same; the $100–$200 tiers exist because agentic coding burns usage fast.

Google is the strongest of the hyperscalers for code, mainly on context — 1M+ tokens means a large share of a real repository fits in the window, which matters more in practice than small benchmark gaps. Sources disagree on whether the current ceiling is 1M or 2M; I couldn’t pin that to a Google page, so plan on 1M.

Microsoft wins one narrow case well: if your code lives in GitHub, Copilot’s integration with pull requests, issues and Actions beats a chat window. Its context is roughly 128K–400K, so whole-repo reasoning isn’t its game.

Not for development: Meta AI (reviews say so plainly — it’s built for Instagram, WhatsApp and Ray-Bans), Alexa+ (a home assistant), Apple Intelligence (on-device convenience features, not a coding tool).

For data processing

Context window decides this, and it’s the spec worth checking before you pay. Google’s 1M+ handles document sets, long logs and large CSVs that Copilot will truncate. Claude and ChatGPT both handle long-document analysis well, with ChatGPT’s 400K context the practical ceiling there.

Microsoft’s advantage is different in kind: it grounds answers in your organisation’s data across SharePoint, Outlook and Teams. No context window substitutes for that if that’s where your data lives.

Where each one is weak

• OpenAI — ten tiers is not a pricing page, it’s a maze, and the $500 tier signals that heavy agentic use is expensive and unpredictable.
• Anthropic — narrower product surface: no image generation, no video, thinner consumer features. You’re buying reasoning and code, not a suite.
• Google — strong capability, but the concentration is the risk: search, email, phone, cloud and assistant become one vendor. Cutting Ultra from $249.99 to $99.99 also suggests pricing isn’t settled.
• Microsoft — the thinnest claim to independence: Copilot has leaned heavily on OpenAI models, so “switch to Copilot to avoid OpenAI” doesn’t fully work. Its own MAI line is about a year old. Retiring Copilot Pro mid-2026 is the kind of churn that breaks workflows.
• xAI — most expensive standard tier at $30, and the weakest enterprise and compliance story of the eight.
• Meta — Llama 4 is a generation off the pace and the 2026 frontier work went to the unreleased Muse line, so the product isn’t running their best. Weakest here for code and long documents.
• Amazon — Alexa+ is a home assistant that happens to be conversational. Free with Prime is excellent value; it is not a work tool.
• Apple — nothing to buy, nothing to cancel, and correspondingly limited. On-device privacy is the genuine advantage; capability is not.

My take

For work, Claude or ChatGPT at $20, and if your budget is the constraint, Gemini AI Plus at $4.99 is the best value in this entire table — a frontier-class model with a 1M-token window for a quarter of the standard price.

Add Microsoft only if your organisation lives in M365, where the grounding genuinely wins. Amazon and Apple are not competing for this work and shouldn’t be evaluated as if they were. Meta and xAI are the two I’d skip for development specifically.

One structural note worth more than any price here: all eight are American. If your concern is a single vendor dropping you, these eight hedge each other nicely. If it’s a jurisdiction, a payment rail, or a policy shift that moves several US companies at once, this page is one basket — which is what the alternatives article is for.

Further reading

• ChatGPT pricing, every plan · ChatGPT vs Claude, tier by tier
• Google AI subscription tiers — primary source
• Microsoft Copilot review · Copilot vs Gemini context and cost
• Grok pricing
• Meta One launch
• Alexa+ free for Prime members — primary source
• Four-way subscription price comparison