Evolving Together

June 2026 · by the Merciful team

When we started Merciful, we had a simple idea: AI chat should be free, accessible, and no-strings-attached. No signup wall, no data harvesting. Just talk!

That core promise isn't changing. But we've learned a lot about what it takes to keep something free and sustainable, and today we're making some changes we want to be transparent about.

What's Changing

Token Limits & Tiers

We're introducing a token-based system to keep the service free for everyone. Here's how it works:

• Anonymous & Unverified users receive a daily grant of 10,000 tokens, enough for meaningful conversation throughout the day.
• Verified Free accounts get 50,000 tokens daily, with the ability to accrue up to 150,000, and access to the Advanced model.
• When you run low, we'll smoothly transition you to our Economy model — a faster, lighter AI that's still helpful, just not as capable. Think of it as a safety net: you're never cut off, just shifted to a more sustainable option.

Model Tiering

Conversations now intelligently shift between models based on context and usage. There are three kinds of model tiers:

• Advanced: Highly capable model for complex reasoning. Only available to registered users, and only for the first few messages in a conversation due to server load.
• Standard: Our balanced model for everyday tasks.
• Economic: A faster, lighter model that is always available.

This transition happens automatically and transparently — you'll see which model is responding right in the UI.

Account Matrix

Here is a breakdown of what each account type currently receives:

We've also added a progress indicator and token counter, so you always know where you stand. No surprises!

Why We're Doing This

Running AI models costs real money. Every message you send costs us tokens on GPU infrastructure. As we've grown, that cost has grown too. We looked at lots of options and are choosing the path that keeps Merciful free for everyone while being honest about the economics. Token limits and model tiering let us serve more people, more sustainably, without compromising our values.

The Technical Side

For the curious, here's some of what changed under the hood:

• Our prompt system now tracks token usage per-user, with daily grants and accrual caps
• Conversations automatically downgrade from our primary model to economy after a threshold
• AI completions are no longer crowdsourced — we now operate all our own infrastructure

This also means we can more easily add new models and backends as they become available.

What's Next

This is just the beginning! Here's what we're working on:

• Context controls — More control over how much conversation history the AI remembers, letting you balance depth and token cost
• Personas & custom prompts — Set a personality or role for your AI conversations
• Image generation — We're exploring bringing image creation to Merciful, powered by the same token system
• Desktop app — A native app experience is on our roadmap
• File editing — Upload and work with documents directly in chat

Your Feedback Matters

The economy model, the token limits, the UI changes — all of these are things we want to get right, and we need your help. If something feels off, if a model gives a bad answer, if the token counts don't make sense, or if you just want to say hi — use the feedback button. We read every single message.

Merciful exists because people like you use it and care about it. These changes are designed to keep it that way.

— The Merciful Team