Gemini Wins 9.1M Users [2026]

Nadia Dubois
October 18, 2026
22 min read

Three AI chatbot apps now sit on almost every Australian’s phone by default: Microsoft Copilot inside Windows and Microsoft 365, Google Gemini inside Android and Workspace, and Meta AI inside WhatsApp, Instagram, Facebook and Messenger. None of them need a download decision the way ChatGPT or Claude do — they are already pinned to a taskbar, a side panel or a chat thread. That default placement is exactly why the comparison matters: most people are not choosing a free AI chat app so much as discovering which one they already have.

Pricing, voice features and the model doing the actual thinking behind each app have all shifted in the past few months. Microsoft has published detailed Australian dollar pricing for Copilot Voice and Microsoft 365 tiers. Google has restructured Gemini’s Australian subscription ladder around usage multipliers rather than flat message caps. And Meta has quietly retired Llama as the brand behind its assistant, replacing it with a new “Muse” model family that now out-scores some rivals on independent benchmarks. This comparison lines up Microsoft Copilot, Google Gemini and Meta AI on pricing, specs, benchmark data and real Australian usage numbers, as of October 18, 2026.

Don’t miss new tech stories on Google

Add Tech Insider once in the Google app and our stories appear in your news suggestions.

Why Comparing Copilot, Gemini and Meta AI Matters Right Now

Most “free AI chatbot” comparisons focus on ChatGPT, Claude and Gemini as competing destinations you actively open. Microsoft Copilot and Meta AI break that pattern. Copilot rides along inside Windows 11, Edge, Word, Excel and Teams, so a huge share of its usage happens because someone clicked a sidebar icon, not because they searched for an AI chat app. Meta AI shows up uninvited in a WhatsApp group chat or an Instagram DM search bar. Gemini sits closer to the traditional model — a standalone app and website — but it is also baked into Android’s assistant button on most new phones sold in Australia.

That distribution advantage is why this particular three-way matchup is more commercially relevant to Australians than a head-to-head between standalone chat apps. If you own a Windows laptop, an Android phone and a WhatsApp account — a combination that covers a large share of the population — you already have access to all three assistants without installing anything. The real question is not “which AI chat app should I download,” it is “which of the AI chat apps already on my devices is worth actually using, and for what.”

There is also a model-level story running underneath the branding. Microsoft leans on OpenAI’s model lineup without publicly naming which exact model answers free Copilot Chat prompts in Australia. Google has shifted its Gemini app between Gemini 3 Pro, Gemini 3 Flash and Gemini 3.5 Flash depending on load and plan, while a new Gemini 4 Argon generation is already topping independent leaderboards elsewhere. Meta has gone further than either, abandoning the Llama name entirely in favour of Muse Spark and Muse Wonder models that Meta’s own research team says are “the most advanced for powering AI assistants” to date. None of these shifts are cosmetic — they change what each free ai chatbot can actually do for an Australian user in October 2026.

Microsoft Copilot in Australia: Plans, Pricing and Voice

Microsoft Copilot’s free consumer tier remains open to anyone with a Microsoft account, but Microsoft is explicit that free usage is “subject to capacity and usage limits,” with the option to upgrade or simply wait and return once a limit resets. That framing — capacity-based rather than a hard daily message count — mirrors the direction most major AI chat apps have taken this year, trading predictable caps for dynamic throttling tied to server load.

Beyond the free tier, Microsoft sells three consumer subscription levels in Australia. Microsoft 365 Personal costs AU$16 per month (or AU$159 billed annually), covers one person across up to five devices, and includes 1TB of cloud storage alongside what Microsoft calls “advanced Copilot” — higher usage limits than the free tier. Microsoft 365 Family runs AU$18 per month and extends the same storage and Copilot access to up to six people, though the AI usage allowance is tied to the subscription owner rather than split evenly. Microsoft 365 Premium, the newest and most expensive consumer tier, costs AU$33 per month and is positioned above Personal and Family with the highest usage ceiling in the consumer lineup.

Voice is where Copilot has pushed hardest in 2026. According to Microsoft’s October 2026 support documentation, Copilot Voice now spans personal accounts (Personal, Family, Premium, Pro), work and school accounts, and even the free Copilot Chat experience — though exact eligibility still depends on subscription tier and region. Activating it is simple: a voice icon sits in the message composer, and tapping it prompts for microphone access. For Australians who want to dictate an email draft or ask a quick question hands-free, that puts Copilot’s voice feature on nearly equal footing with Gemini’s long-running voice mode, at least on paper.

On the business side, Copilot Chat for eligible work accounts connects to an organisation’s Microsoft 365 data through what Microsoft calls Work IQ, an intelligence layer that reads a tenant’s documents, emails and chats to personalise answers. That enterprise capability is a different product from the ordinary consumer Copilot app most Australians use on a personal laptop, and the two should not be confused when comparing free-tier experiences.

Google Gemini in Australia: Free Tier, Paid Plans and Gemini 3 Pro

Gemini’s Australian pricing page lists a free tier at A$0 per month, available to anyone signed in with a Google Account. Above that, Google now runs a tiered usage-multiplier system rather than simple message counts: a A$7.99 per month plan doubles the free allowance, and a A$32.99 per month Google AI Pro plan quadruples it. At the top end, Google AI Ultra starts at A$149.99 per month for five times the usage of AI Pro, scaling up to A$329.99 per month for a tier offering twenty times AI Pro’s usage — aimed squarely at power users running heavy image generation or long research sessions. Google’s global updates page separately notes Google AI Ultra “starting at $100 USD/month” in some markets, underlining that Australian consumer pricing on Gemini’s own local subscription page is the figure to trust for local budgeting, not a US dollar conversion.

Usage limits inside the Gemini app are calculated dynamically rather than fixed. Google states that limits depend on prompt complexity, which features are in use, and overall chat length, with access refreshing roughly every five hours and an additional weekly ceiling layered on top. In practice, that means two free Gemini users asking very different kinds of questions can hit very different effective limits in the same day — a short factual question costs less “usage” than a long multi-turn research session or an image generation request.

Google’s Australian Gemini page explicitly names Gemini 3 Pro as available to users, alongside Nano Banana Pro for image generation and a Deep Search mode inside AI Mode, with access varying by country and plan. That is a meaningfully different situation from Meta AI, where no equivalent official page names the exact model build behind the consumer-facing assistant. It is worth noting that a newer Gemini 4 Argon generation has already appeared on independent leaderboards as Google’s most capable model, but it is not yet confirmed as the model answering everyday prompts inside the free or AI Pro consumer Gemini app in Australia — a distinction that matters when judging what you are actually getting for A$0, A$7.99 or A$32.99 a month.

Meta AI in Australia: From Llama to the Muse Model Family

The biggest change behind Meta AI in 2026 has nothing to do with the app’s interface and everything to do with what powers it. Meta has walked away from the Llama brand that defined its models since 2023, replacing it with a new “Muse” family. Meta’s own AI Research blog documents the cadence: Muse Spark 1.2 landed on August 5, 2026, Muse Glimmer 30B followed on August 10, Muse Voice Transcribe arrived September 1, Muse Spark 1.3 shipped September 2, and the flagship Muse Wonder and Muse Wonder Full models launched October 12, 2026, with an open-weight Muse Spark Open release on October 15. That is eight model releases in roughly three months — a release cadence independent tracker Opper.ai describes as “a new release typically every 25 days” over the preceding twelve months.

Meta’s framing of Muse Wonder is ambitious. The company’s October 12 announcement describes the new models as its “most capable yet for personal and professional agentic tasks, with stronger memory, focus, coding, and safety,” and positions them as rivalling OpenAI and Anthropic’s frontier releases. Independent benchmark data backs some of that claim up: Artificial Analysis Intelligence Index v4.3 scores Muse Wonder Full at 56 (roughly 55.8%), placing it in the same tier as Claude Sonnet 5.5 and Claude Opus 5.5, and ahead of GPT-6 Astra and Gemini 4 Argon, which both sit at 53 on the same index according to a benchmark summary published by officechai.com.

Where the comparison gets harder is on the consumer side. Unlike Microsoft and Google, Meta has not published an Australian consumer price for the Meta AI assistant inside WhatsApp, Instagram, Facebook or Messenger, nor has it disclosed a specific context window, an official voice-mode brand name, or a current monthly active user figure for that specific product. Reporting around the Muse Wonder launch from Euronews notes that accessing the underlying Wonder models through Meta’s developer-facing API “could incur a monthly subscription ranging from $20 to $100,” but that figure applies to API and business access, not to the free consumer chatbot embedded in Meta’s social apps. For everyday Australian users, Meta AI functions as a free assistant with no confirmed local subscription tier — a transparency gap that stands out next to Microsoft’s and Google’s publicly itemised Australian dollar pricing.

Full Specs Comparison: Copilot vs Gemini vs Meta AI

The table below lines up verified specifications across all three apps. Where a figure has not been publicly disclosed by the company — which happens most often with Meta AI — that is noted explicitly rather than estimated.

Pricing Comparison: What You Actually Pay Each Month

Stripped of feature detail, the pricing ladders for Copilot and Gemini look more alike than the branding suggests: both offer a genuinely free tier, a mid-priced individual plan in the AU$16–$33 range, and in Gemini’s case a much steeper top tier for heavy users. Meta AI simply does not appear in this kind of table with real numbers, because Meta has not published one for its consumer assistant.

The practical takeaway for an Australian household budgeting for a free ai chatbot habit: Gemini’s free tier is unconditionally free with no Microsoft account or Microsoft 365 purchase required, while Copilot’s most useful features increasingly sit behind a Microsoft 365 subscription that most people already pay for storage and Office apps anyway. Meta AI remains the only one of the three with zero visible price tag anywhere in its consumer product, for better or worse.

Benchmark Performance: How the Underlying Models Rank

Benchmark scores do not map perfectly onto “which free ai chatbot answers my question best,” because none of the three companies has confirmed exactly which model build serves every free-tier prompt. Still, the benchmark data on the model families involved is informative, and it comes from three independent sources that track AI models separately: the Artificial Analysis Intelligence Index, the LMArena (Chatbot Arena) Elo leaderboard, and Artificial Analysis’s separate Coding Agents Index and Terminal-Bench 4.0 results.

On the Artificial Analysis Intelligence Index v4.3, dated October 16–17, 2026, OpenAI’s GPT-5.6 Sol currently leads the entire leaderboard at 58.9%, followed by Claude Opus 5.5 at 57.6% and Claude Sonnet 5.5 at 56.0%. Meta’s Muse Wonder Full sits close behind at 55.8% (rounded to 56), tying it with Claude Sonnet 5.5 in the same performance band, while the smaller Muse Wonder scores 52.4%. Google’s Gemini 3.5 Flash scores 50.2% on the same index, and an estimated figure for Gemini 3 Pro Deep Think sits near 46.9%. Because GPT-5.6 Sol is OpenAI’s model and Microsoft licenses OpenAI technology for Copilot, that top score is suggestive of Copilot’s potential ceiling — but Microsoft has not confirmed GPT-5.6 Sol specifically answers free consumer Copilot Chat prompts in Australia, so it should be read as “the best OpenAI model available,” not as a confirmed Copilot benchmark.

On the LMArena Text Arena Overall leaderboard, using board data through October 15–18, 2026, Google’s newer Gemini 4 Argon generation leads with a preliminary score of 1,524±9 Elo, with Meta’s Muse Wonder (Max) close behind at 1,509±11 — a gap of just 15 points. Further down the same board, Gemini 3.8 Flash (high) scores 1,496±5, Gemini 3.7 Flash (high) scores 1,488±5, and the Gemini 3 Pro model named on Google’s Australian consumer product page scores 1,485±4. That spread illustrates a real gap between Google’s newest, most capable model (Gemini 4 Argon) and the model actually named on the consumer Gemini app’s Australian product page (Gemini 3 Pro) as of mid-October 2026.

A third data point comes from Artificial Analysis’s specialised coding benchmarks. Analyst summaries of Muse Wonder’s release note that Meta’s Terminal-Bench 4.0 score jumped from 33% for Muse Spark 1.3 to 57% for Muse Wonder Full, and that Muse Wonder Full’s Coding Agents Index score of 68 ties it with the Claude Code plus Claude Sonnet combination — a sharp six-week improvement Artificial Analysis itself describes as “a material jump in intelligence over previous Meta models.” None of this is directly visible to a casual WhatsApp user asking Meta AI for a recipe, but it explains why Meta’s model team has been able to close so much ground on OpenAI, Google and Anthropic in such a short window.

Who Is Actually Using Each Chatbot in Australia

Two independent Australian research firms have published usage figures for these apps in 2026, and their numbers do not fully agree — a reminder that “how many people use this” is measured differently depending on survey methodology, timing and age range.

Roy Morgan Research reported that “Google Gemini was used by 5 million Australians, equivalent to 21% of Australians aged 14 and over, in the March quarter of 2026,” and separately found that “Microsoft Copilot was used by an estimated 4 million Australians, equivalent to 17% of Australians aged 14 and over,” in the same quarter. Roy Morgan’s age breakdown adds useful texture: “Among Australians aged 14 and over, Microsoft Copilot usage was highest among people aged 35–49 at 24%, followed by 21% among those aged 25–34 and 18% among those aged 50–64” — a pattern consistent with Copilot’s heavy presence in workplace laptops rather than personal devices skewing younger.

Telsyte, publishing its own Australian Artificial Intelligence Study 2026, reported substantially higher figures across the board: “Telsyte reported 9.1 million Australian users for Google Gemini, 5.6 million for Meta AI, and 5.4 million for Microsoft Copilot.” Telsyte’s broader finding puts the scale of AI adoption in context, estimating that “17.4 million Australians, or 77% of the population aged 16 and over, used artificial intelligence during the study period.”

Both data sets agree on the relative order: Gemini ahead of Copilot, with Meta AI landing in between the two in Telsyte’s figures (Telsyte is the only firm of the two that published a Meta AI number at all). The gap between Roy Morgan’s and Telsyte’s absolute figures — 5 million versus 9.1 million for Gemini, for instance — likely reflects different survey windows, sample sizes and how each firm defines “used,” but the consistent ranking across two independent Australian research houses is the more durable signal for anyone trying to gauge real-world reach rather than marketing claims.

Voice Mode and Multimodal Features Compared

Voice has become the feature all three companies are racing to polish, but the maturity gap between them is still visible. Google’s voice conversations in Gemini are the most established of the three, built on several years of Google Assistant infrastructure and now layered with Gemini’s language model for more natural back-and-forth exchanges, including interruption handling and follow-up questions without restarting the conversation.

Microsoft’s Copilot Voice is newer but has expanded quickly in 2026, now reaching personal, Family, Premium, Pro, work, school and even free Copilot Chat tiers according to Microsoft’s own October 2026 support documentation — though the company is careful to note that “voice availability depends on your subscription and region,” so an Australian free-tier user should not assume parity with a paid US subscriber. In practice, Copilot Voice works best for short, task-focused exchanges: dictating a document outline, asking Copilot to summarise an open Excel sheet, or getting a quick definition while working inside a Microsoft 365 app.

Meta AI’s voice story is the least documented of the three for ordinary consumers. Meta’s research division has shipped a model called Muse Voice Transcribe, released September 1, 2026, but the available evidence does not confirm that this specific model powers a consumer-facing voice mode inside the WhatsApp or Instagram version of Meta AI, nor does it establish what that feature is branded as for end users. Anyone relying on Meta AI for voice interaction in Australia should test it directly inside their own WhatsApp or Instagram app rather than assume feature parity with Copilot or Gemini, since Meta has not published the same kind of detailed support documentation Microsoft and Google maintain.

On image generation, Gemini’s Nano Banana Pro is explicitly named on Google’s Australian product page as part of the consumer offering, giving Gemini the clearest documented image-generation story of the three. Copilot’s image tools lean on Microsoft Designer integration inside Microsoft 365 apps, useful for quick slide or document graphics but less of a standalone creative tool. Meta AI’s image generation is woven directly into Instagram and Facebook’s posting and messaging flows, which makes it convenient for social content but harder to use as a general-purpose image generator outside those apps.

Real-World Use Cases: Five Ways Australians Are Actually Using These Apps

These scenarios illustrate how the three apps tend to get used in practice, based on where each one is physically placed inside an Australian user’s existing apps and devices.

• Small business invoicing and quotes: A Sydney tradesperson running Microsoft 365 Personal for AU$16/month uses Copilot inside Word and Excel to draft job quotes and tidy up a pricing spreadsheet, treating the AI feature as a bonus on top of software they already pay for rather than a separate product decision.
• University research and assignments: A Melbourne student on Gemini’s free A$0 tier uses it for research summaries and leans on Nano Banana Pro to generate quick diagrams for assignments, upgrading to the A$7.99 tier only during exam periods when the five-hour usage window starts to feel restrictive.
• Customer replies on social media: A Perth retailer uses Meta AI inside their business WhatsApp and Instagram accounts to draft quick replies to customer questions about stock and store hours, taking advantage of the fact that the assistant is already sitting inside the messaging app they use for sales anyway.
• Enterprise rollout evaluation: An IT manager at a mid-sized Australian firm compares Copilot Chat’s Work IQ layer against Google Workspace’s Gemini integration for a 200-seat deployment, weighing Microsoft’s deeper Office document integration against Google’s lower per-seat Workspace pricing.
• Hands-free queries while driving: A regional sales rep uses Gemini’s voice mode through Android Auto for quick fact-checks and scheduling questions during long drives between country towns, favouring Gemini’s more mature voice handling over Copilot Voice’s narrower regional availability.

Migration Guide: Moving Between Copilot, Gemini and Meta AI

Switching your default AI chat app, or running two in parallel, is less disruptive than switching between most other software categories, because none of the three apps lock conversation history behind an exclusive export format the way some dedicated chatbot platforms do. The steps below work for most Australian consumer accounts.

1. Export or screenshot any Copilot conversations you want to keep before downgrading a Microsoft 365 subscription, since advanced Copilot usage limits drop immediately once you move from Premium or Personal back to the free tier.
2. Use Google’s Takeout tool to export Gemini conversation history if you plan to close or downgrade a Google AI Pro or Ultra subscription, since usage limits reset the moment your plan changes.
3. Check Meta’s Accounts Center data download tools for any Meta AI chat history inside WhatsApp or Instagram, since there is no dedicated “export Meta AI chats” button separate from the general data download process.
4. Re-create any saved prompt templates manually, since none of the three apps currently support importing a competitor’s saved prompts or custom instructions directly.
5. Match the new app to the task rather than trying to fully replace one with another: move document-heavy work to whichever app sits inside your office suite, move quick social replies to Meta AI if that is where the conversation already is, and move voice-first tasks to Gemini given its more mature voice handling.
6. Reconfirm your monthly budget against the pricing table above before committing to a paid tier, since Google AI Ultra’s A$149.99–A$329.99 range sits well above Microsoft’s AU$33 ceiling for consumer Copilot.
7. Test voice mode directly on your own device for all three apps rather than assuming feature parity, since Microsoft explicitly ties Copilot Voice availability to “subscription and region,” and Meta has not published equivalent documentation for its consumer voice feature at all.

A simple way to standardise how you prompt each assistant during a transition is to keep one plain-text instruction block and paste it into whichever app you are testing that week:

Context: Australian small business, GST-registered, replying to a customer enquiry.
Task: Draft a short, polite reply confirming stock availability and estimated delivery time.
Tone: Friendly, concise, no more than 80 words.
Constraint: Do not quote a price unless one is provided in the conversation.

Running the same instruction block through Copilot, Gemini and Meta AI side by side is the fastest way to judge which free ai chatbot actually fits a specific recurring task, rather than relying on benchmark scores that measure very different kinds of reasoning.

Pros and Cons of Each AI Chat App

Microsoft Copilot

• Pros: Deep integration with Word, Excel and Teams; clear Australian dollar pricing across three consumer tiers; Copilot Voice now reaches most tiers including free Copilot Chat; Work IQ adds genuine value for business users already on Microsoft 365.
• Cons: Free tier capacity limits are not published as specific numbers; Microsoft has not confirmed which exact OpenAI model answers free consumer prompts in Australia; no consumer tier above AU$33/month for users who want the highest usage ceiling.

Google Gemini

• Pros: Genuinely free A$0 tier with no bundled subscription required; Gemini 3 Pro and Nano Banana Pro explicitly named on the Australian product page; mature, well-documented voice mode; widest consumer pricing range for users who want more usage headroom.
• Cons: Usage limits are dynamic and hard to predict in advance; top Google AI Ultra tier is expensive relative to Copilot and most AU budgets; the newer, higher-scoring Gemini 4 Argon model is not yet confirmed as what free-tier users actually receive.

Meta AI

• Pros: Already embedded inside WhatsApp, Instagram, Facebook and Messenger with no separate app needed; underlying Muse Wonder Full model benchmarks competitively against Claude and Gemini on the Artificial Analysis Intelligence Index; no visible price tag for everyday consumer use.
• Cons: No published Australian consumer pricing, context window, or voice-mode branding; no confirmed monthly active user figure from Meta itself; least transparent of the three on exactly which model answers a given prompt at a given time.

Privacy and Data Handling Considerations

All three companies process conversation data to operate their respective assistants, and all three sit under different existing privacy frameworks depending on which other products a user already has connected. Microsoft’s consumer Copilot sits inside the same Microsoft account ecosystem as Outlook, OneDrive and Windows sign-in, meaning conversation data potentially sits alongside other personal files depending on account settings. Google’s Gemini similarly ties into a broader Google Account, with conversation history manageable through the same activity controls Google already provides for Search and Maps history.

Meta AI carries a distinct consideration because it operates inside messaging apps that many Australians already treat as more private by default, particularly WhatsApp with its end-to-end encrypted messaging for human-to-human chats. Conversations directed at Meta AI itself are not covered by that same end-to-end encryption, since the assistant needs to read the message to respond — a distinction that is easy for a casual user to miss when Meta AI appears inside the same chat window as an encrypted conversation with a friend. Anyone using Meta AI for anything sensitive should treat it the same way they would treat any other cloud-based assistant, not as an extension of WhatsApp’s encrypted messaging guarantees.

Which Chatbot Should You Choose: Five Use-Case Recommendations

• Already paying for Microsoft 365: Use Copilot first. The advanced usage limits and Work IQ features on Personal, Family or Premium are effectively bundled into a subscription most households already carry for Office and OneDrive storage.
• Budget-conscious, no existing subscriptions: Use Gemini’s free A$0 tier. It requires nothing beyond an existing Google Account and offers the clearest documented free feature set of the three, including Gemini 3 Pro access and Nano Banana Pro image generation.
• Small business customer service on social media: Use Meta AI. Its presence directly inside WhatsApp Business and Instagram DMs removes the friction of asking customers to use a separate app, even without a published consumer price to compare against.
• Heavy daily use, research or long documents: Consider Google AI Pro at A$32.99/month for the 4x usage multiplier before jumping straight to the much pricier Ultra tiers, since most individual power users will not need the 20x ceiling that A$329.99/month buys.
• Enterprise deployment across a whole organisation: Pilot both Copilot Chat with Work IQ and Google Workspace’s Gemini integration against the same real workflows before committing, since the two platforms integrate with very different document and email ecosystems that are expensive to migrate later.

The Verdict: Copilot vs Gemini vs Meta AI

There is no single winner across all three apps, because they are not really competing for the same moment of use. Gemini wins on transparency and documented free-tier features, with a genuinely free A$0 plan, named model access to Gemini 3 Pro, and the most mature voice mode of the three based on published documentation. Copilot wins for anyone already inside the Microsoft 365 ecosystem, where its document and spreadsheet integration turns an AI chat feature into a natural extension of software people already pay for, backed by clear AU$16–$33 monthly pricing. Meta AI wins purely on distribution and benchmark upside: Telsyte’s 5.6 million user figure and Muse Wonder Full’s 56-point Artificial Analysis Intelligence Index score (ahead of GPT-6 Astra and Gemini 4 Argon, both at 53) suggest Meta’s underlying technology has closed real ground, even though the consumer product around it remains the least documented of the three.

For most Australians, the practical answer is not to pick exactly one. Roy Morgan’s and Telsyte’s usage figures both show large, overlapping user bases rather than a single dominant app, which matches how these three assistants actually behave in daily life: already installed, already free at the entry level, and each strongest in the specific app or device where it already lives.

Frequently Asked Questions

Which AI chatbot app is completely free in Australia?

Google Gemini has the clearest unconditionally free tier at A$0/month, requiring only a Google Account. Microsoft Copilot’s free consumer tier exists but is capacity-limited without a specific published number. Meta AI appears free for everyday consumer use with no confirmed Australian subscription tier at all.

Does Microsoft Copilot or Google Gemini have a better voice mode?

Google Gemini’s voice mode is more established, built on years of Google Assistant infrastructure. Copilot Voice has expanded rapidly in 2026 to reach most subscription tiers including free Copilot Chat, but Microsoft itself notes that availability depends on subscription and region, so it should be tested directly rather than assumed.

What AI model actually powers Meta AI now that Llama is gone?

Meta has replaced Llama with a new Muse model family throughout 2026, culminating in Muse Wonder and Muse Wonder Full launched October 12, 2026. These score 52 and 56 respectively on the Artificial Analysis Intelligence Index. Meta has not confirmed exactly which Muse model build answers every consumer Meta AI prompt inside WhatsApp or Instagram.

Is Microsoft Copilot’s free Australian plan different from the paid Microsoft 365 plans?

Yes. The free tier uses capacity-based limits without a published number, while Microsoft 365 Personal (AU$16/month), Family (AU$18/month) and Premium (AU$33/month) all include what Microsoft calls “advanced Copilot” with higher usage ceilings than the free experience.

How many Australians use each AI chatbot app?

Roy Morgan reported 5 million Australians using Gemini and 4 million using Copilot in the March 2026 quarter. Telsyte’s separate Australian Artificial Intelligence Study 2026 reported higher figures: 9.1 million for Gemini, 5.6 million for Meta AI, and 5.4 million for Copilot. The two firms disagree on absolute numbers but agree that Gemini has the largest reach of the three.

Can I use Gemini, Copilot and Meta AI together, or do I need to pick one?

There is no technical reason to pick just one. All three are already free to access at the entry level and sit inside different apps and devices, so many Australians end up using whichever one is already open: Copilot inside Microsoft 365 documents, Gemini for general search-style questions, and Meta AI inside WhatsApp or Instagram conversations.

Which chatbot app is best for small business customer service in Australia?

Meta AI has a practical edge for businesses that already handle customer enquiries through WhatsApp Business or Instagram DMs, since it sits directly inside those conversations without requiring customers to switch apps. Businesses doing more document-heavy quoting and invoicing may get more value from Copilot’s Microsoft 365 integration instead.

Is my data safe with Meta AI inside WhatsApp and Instagram?

Messages sent to Meta AI are not covered by WhatsApp’s end-to-end encryption that protects human-to-human chats, because the assistant needs to read the message to generate a response. Users should treat conversations with Meta AI the same way they would treat any other cloud-based AI chat app, rather than assuming WhatsApp’s usual encryption guarantees extend to it.