What makes ByteChat different
One chatroom.
Every AI. Your keys.
Most AI apps give you one model in a private chat. ByteChat is a multi-bot chatroom where Claude, GPT, Gemini and more answer together — compare them side by side, chain their work, or let a judge bot weigh every answer into one verdict. All on your own API keys, at provider cost with zero markup. Here's everything you can do, by plan.
Free BYOK + all the essentials
Pro $4.99/mo · power tools
The multi-bot chatroom
These are the things a normal one-model chat app simply can't do. Every tier gets the core room — paid tiers fuel it with real API keys.
💬
Many AIs in one conversation
FreePro
Add multiple bots to a single chatroom. Broadcast a message to all of them at once, or address one with an @mention. Their replies stream in concurrently.
Why it's unique: a normal AI app is a 1-on-1 thread. Here you can put Claude, GPT, Gemini, Grok, DeepSeek and more in the same room and compare or combine their answers without copy-pasting between tabs.
- Add a bot — tap + Add bot, paste your API key (we auto-detect the provider) and pick a model.
- Repeat for as many bots as you like — unlimited on every plan.
- Type a message — by default it goes to every bot in the room.
- Or target one — start with @BotName to ask just that bot.
🔑
Bring Your Own Key — zero markup
FreePro
Paste your own provider API keys. ByteChat sends requests straight to the provider — your key never touches our servers beyond the direct proxy, and we add no token markup.
Why it's unique: instead of stacking $20/mo subscriptions per model, you connect your own keys for free and pay only your real (often pennies) usage cost — across every provider, with no token markup.
- Open Settings → API Keys.
- Paste a key for any provider (Claude, GPT, Gemini, Groq, DeepSeek and more).
- Keys are stored locally in your browser only — never on our servers.
- Start chatting — usage is billed by your provider at their rates.
🎭
Per-bot personas & room identity
FreePro
Give each bot its own system prompt — Critic, Builder, Analyst — and set a shared chatroom context that every bot sees.
Why it's unique: you design a panel of specialists in one room instead of re-pasting instructions into separate chats.
- Tap a bot to edit it, then set its system prompt / role.
- Set room context from the chatroom settings to brief every bot at once.
📤
Side-by-side panels & export
FreePro
Switch to panel view to read each bot's answer in its own column, then export the conversation to Markdown, HTML or PDF — on every plan.
- Toggle panel view to compare replies column-by-column.
- Tap Export and choose your format.
🔗
Chain Mode
Pro
Make your bots reply in sequence instead of in parallel, in an order you choose. Each bot reads the previous bots' answers, so you can run a pipeline — one drafts, the next critiques, the third rewrites — from a single message.
Why it's unique: turns a room of independent bots into a collaborative assembly line you arrange yourself, with no manual copy-pasting between models.
- Have 2+ bots active in the room.
- Tap the chain icon (next to the mic) to switch on chain mode — the Chain order panel opens.
- Set the sequence with the up/down arrows (each bot shows its role), then close the panel.
- Send your message — bots reply one after another in your order, each building on the last.
- Tap the chain icon again to reorder anytime, or turn chain mode off from inside the panel.
Pro power features
Free gives you every AI. Pro makes them work together — and work for you.
⚖️
Consensus / Judge mode
Pro
Ask every bot at once, then a judge bot reads all the answers and gives you a single verdict — what they agree on, where they disagree, and a confidence score. Stop guessing which AI is right.
Why it's unique: a single-model app gives you one opinion you have to trust blindly. ByteChat turns several models into one reviewed answer — with the disagreements surfaced, not hidden.
- Have 2+ bots active in the room.
- Open the mode menu (the 🧭 icon) and pick Consensus.
- Ask your question — every bot answers, then the verdict card appears below with a confidence score.
🧭
Smart model routing
Pro
Instead of broadcasting, ByteChat sends each question to the cheapest model that can handle it — simple asks to a fast cheap model, hard ones to a frontier model — and shows you what you saved each time.
Why it's unique: it turns your stack of models into one cost-aware assistant. Pay less on everyday questions without thinking about which model to pick — and escalate to the strongest one in a single tap when you want to.
- Open the mode menu and pick Smart routing.
- Ask normally — ByteChat picks the right model and shows the saving vs your priciest bot.
- Tap "Ask … instead" on the routing note to re-run with your strongest model if you want more.
⬢
Elevated Thinking
Pro
A 4-stage reasoning protocol from a single question. Your AIs frame the problem independently, then their ideas are pooled and de-identified, pressure-tested across the whole set, and synthesised — so you get genuinely combined thinking, not just parallel answers.
Why it's unique: it removes the two things that quietly wreck multi-AI output — early anchoring (models copying each other) and false consensus (everyone agreeing to be agreeable) — by keeping the first pass blind and explicitly preserving real disagreement.
- Have 2+ bots active in the room.
- Tap the layers icon (next to the chain icon) to switch on Elevated Thinking.
- Stage 1 — Independent framings: each AI frames the problem blind, without seeing the others.
- Stage 2 — Pooled & de-identified: all ideas merge into one anonymous candidate space.
- Stage 3 — Lateral pressure: each AI merges overlaps, exposes assumptions, and revises openly.
- Stage 4 — Synthesis: each AI writes one summary that keeps the real tensions instead of forcing consensus.
- Stage 5 — Your call: read the syntheses and pick or braid them forward. (Runs ~4× the API calls of a normal message.)
🌿
Conversation Branching
Pro
Fork any conversation at any message into a new branch. The original stays untouched, so you can explore an alternative direction without losing your thread. The parent shows how many branches grew from it.
Why it's unique: explore "what if" paths the way you'd branch code — instead of overwriting or cluttering one long thread.
- Tap any message (yours or a bot's) to reveal its action bar.
- Tap "Branch" — a new conversation is created with everything up to that point.
- Keep chatting in the branch; the original is preserved.
- Spot branches in the sidebar by the branch tag and count.
📊
API Cost Analytics Dashboard
Pro
Pro extends the Savings Tracker into a full dashboard: a 12-month spend bar chart, up to 11 months of history, a monthly budget alert with progress bar, and one-tap CSV export of all your usage.
Why it's unique: treat your own-key AI spend like a real budget line — track trends, set a cap, export for expenses.
- Open the Usage & Savings panel — Pro shows the 12-month chart automatically.
- Set a monthly budget in the field provided to get a progress bar and alert.
- Tap "Export CSV" to download your full usage history.
🎙️
Voice-to-multi-bot
Pro
Speak once and your words go to every bot in the room at the same time — no typing, no repeating yourself per model.
- Tap the mic and talk.
- Send — all active bots hear the same prompt.
🔊
Voice reply
Pro
Have any bot's answer read aloud — and because you run several bots, each one gets its own distinct voice, so a multi-bot conversation actually sounds like different speakers.
Why it's unique: listen to a chain or consensus discussion hands-free, with each model in its own voice instead of one flat narrator.
- Tap a bot's message to reveal its action row.
- Hit Speak — the answer plays in that bot's assigned voice.
- Tap Speak again to stop.
▶️
Live code execution
Pro
Run code blocks a bot produces right inside the chat — JavaScript, HTML/CSS preview and more — in a sandbox, without leaving ByteChat.
- Ask a bot for code.
- Tap "Run" on the code block to execute and see the output inline.
🗂️
Room Templates
Pro
Save a full room setup — the bots, their roles and the context — then restore it with one tap whenever you need that team again.
- Build your room with the bots and prompts you want.
- Save as a template, then restore it any time for a fresh conversation.
Stack your keys. Keep your savings.
Start free — bring your own keys, run every model, chain your bots — then go Pro to unlock branching, Elevated Thinking, voice and code execution.