Also, (and I request you to not consider this as a defense, but rather a reflection on the entire AI product industry...)

Even the big players conveniently hide usage stats and usage limits under help and support blogs. Not that it's an excuse to do the same at Merlin too, but it shows that no one has been able to crack transparency and layperson communication of usage limits, when it comes to LLM usage.

By comparison, we actually offer equivalent usage to models (and actually higher usage on o4-mini series) on Pro vs ChatGPT Plus.

We're fixing our pricing page to make things more transparent, as well as adding a fair usage tracker visible conveniently within the Merlin UI.

Apologies for the bad experience -- in hindsight, it is something we should have fixed a long while back. But better late than never, I suppose :')

o3 is an extremely costly model, and it is not possible to hand it out at large to users (and we made sure not to communicate that there is unlimited usage for o3 either)

Even OpenAI restricts users to around 10 per day, and that is because they price the API way more than what it costs them.

Gemini 2.5 Pro is actually extremely smart and capable compared to o3 at a great price. I've personally tested it for reasoning and problem solving prompts, and I'd recommend using it for research tasks. You can also consider Claude 3.7 Sonnet (Thinking).

Yeah, the team has been pretty vague, and I've seen many posts about how Merlin is not at all being straightforward about usage and user limits. It is a mjor downside of their service. The stupid, un-labeled and vague "fair limit" bar needs a rework.