Merlin AI alternatives: 11 picks, and Merlin's $100 usage cap

Which Merlin AI alternative fits the reason you are leaving?

A yearly Merlin Pro plan bills $228 up front, which Merlin's pricing page shows as $19 a month as of 3 October 2026. Merlin Teams also shows $19, but that's a price per seat on monthly billing. Every row below is in US dollars, read from the vendor's own page on 3 October 2026 unless the row says otherwise.

Three of these rows are worth a second look. Monica's pricing tooltip promises 1,600 advanced queries a month on Max, and that's a number Merlin never prints for Pro. Claude's plan table puts Claude in Chrome on Pro. So if you mostly asked Claude in Merlin, you can get a browser agent from Anthropic itself for $9 less. Poe's pricing page refused our automated read, so its row comes from a tracker and not from Poe. Monica and ChatHub each have their own full breakdown.

Is Merlin AI unlimited?

No. Merlin's own pricing FAQ answers that question with "Merlin Pro gives atleast 30x the usage of Merlin Free, but is bound by Fair Use Limits", while the plan card above it says 50x. Neither multiple is a number you can plan around, because Merlin Free's allowance isn't printed either. The real ceiling sits in section 12 of Merlin's terms, read on 3 October 2026. A standard paid account whose usage costs Merlin more than $100 in a month loses service for the rest of that month. Past $16 in a day, it loses service until the next day.

Merlin counts those dollars itself, from the tokens your prompts use plus its cloud costs. So we turned the daily $16 into a count of answers you could get for it. We used OpenRouter's list price for Claude Opus 5.5, which was $4 per million input tokens and $20 per million output on 3 October 2026, and the same for GPT 6.1 Sol ($2 and $10). Those are the two "exclusive models" on Merlin's Pro card. Each answer in our count is 500 tokens long. The size of your prompt is what moves the number.

Short chats will never hit it. A long PDF will. Merlin's FAQ puts the Pro context window at 100K tokens, and a prompt that fills it costs about $0.41 on Opus 5.5 at list price. That's 39 of them in a day, before Merlin adds any of its cloud costs. Treat these as upper bounds. We don't know what Merlin pays for its API access, and its own count includes costs that we can't see.

Two more clauses matter. Section 11 lists usage costs "2.5x or more than the monthly fees charged" as grounds for suspension with no refund, which is $72.50 on a $29 plan, or about $47.50 at the $19 yearly rate, and lower than the $100 ceiling. And on discounted plans, both caps shrink "around proportionally" to the price you paid. Merlin's site was running a $5 a month Pro offer on 3 October 2026, so that offer's ceiling is roughly a sixth to a quarter of the figures above.

Is Merlin AI safe to use?

Merlin is a real product from Foyer Tech Inc., but an independent study found it collecting more than it should. UCL's August 2025 study of ten popular AI browser extensions, presented at the USENIX Security Symposium, reported: "One assistant, Merlin, even captured form inputs such as online banking details or health data." The same researchers found Merlin and Sider kept recording when users switched to a private space.

Merlin isn't alone in that study, and its alternatives aren't clean either. The researchers found Monica and Sider inferring age, gender, income and interests to personalise answers. Only Perplexity showed no evidence of profiling. Merlin's pricing FAQ now says it stores all data in the United States under GDPR, SOC2 and ISO standards. DeepSeek's privacy policy says it stores personal data in the People's Republic of China. If privacy is why you're leaving, a sidebar that reads every page is the wrong replacement. Use the labs' own apps, or a workspace you open on purpose.

Can you get a refund when you leave Merlin?

Only within 3 days of purchase. Merlin's refund policy, updated 12 August 2026, covers every subscription bought on or after that date. After day 3 all purchases are final, there are no prorated refunds for unused time, and low or no usage doesn't qualify.

So a yearly subscriber who paid $228 and leaves in month two won't get the rest back. Cancel anyway, because cancelling stops the next renewal. Then save the chats you care about before your access ends. Whizi imports ChatGPT and Claude exports, not Merlin's, so if your history matters, copy the important threads out by hand.

Where does Whizi beat Merlin, and where does it lose?

Whizi, which publishes this page, loses to Merlin on the top models per dollar and wins on a cost you can see before you send. Start with the losses. Merlin Pro at $29 includes Claude Opus 5.5 and GPT 6.1 Sol. Whizi Pro at $29.99 carries Claude Opus too, but at 20 credits a message its allowance covers 125 Opus messages a month. Merlin also takes a 100K token context on Pro, while Whizi gives every model a budget of 40,000 input tokens per turn. If your day lives in Merlin's Gmail, YouTube and page tools, Sider or Monica replaces that more directly than any chat workspace.

The win is predictability. Every Whizi model costs a fixed number of credits per message, and neither the length of your prompt nor the length of the thread changes it. Claude Sonnet is 10 credits, so Pro's 2,500 credits carry 250 Sonnet messages. Powerhouse carries 10,000 credits and 280+ models. That's the opposite of a dollar ceiling that Merlin computes on its side, from costs you can't see. Projects hold up to 10 pinned files and 32,000 characters of instructions each. Always-on Skills apply your standing instructions to every chat. And Whizi imports ChatGPT and Claude exports, so history from either lab comes with you. Pro, where Claude Sonnet, ChatGPT Sol and Kimi K3 start, opens with a $1.99 first week.

Merlin AI vs ChatGPT Plus: which is the better deal?

ChatGPT Plus is the better deal if GPT is the only model you use: it's $20 a month on OpenAI's help center, $9 less than Merlin Pro. Merlin is the better deal if you switch between labs daily and want Claude Opus 5.5 and GPT 6.1 Sol behind one login.

Merlin's FAQ argues Pro offers "5x the context window as ChatGPT Plus, as of April 2025", a comparison that is now 18 months old. The same logic applies to Claude Pro, also $20 in the US. One lab at full depth beats five labs behind a fair use ceiling, unless you really do use all five.