Frequently Asked
Questions

Most users save 70-90%. AI subscriptions add up fast — ChatGPT Plus is $20/mo, Claude Pro is $20/mo, Gemini Advanced is $20/mo, Perplexity Pro is $20/mo. That's $80/mo just to access four providers. With ByteChat + BYOK, typical users spend about $5-12/month all-in — bring-your-own-key is free, so you pay only a few dollars of per-token usage with no markup (optionally add the $4.99/mo Pro plan for power features) — across all the providers they use. Our in-app Subscription Savings Tracker shows your exact savings in real time.

Several providers offer real-time web search via their APIs — you don't have to give that up by switching to BYOK. Perplexity is built for live web search and returns cited sources. Grok (by xAI) can search the web and X/Twitter in real time. Gemini models support Google Search grounding. Add any of these as a bot in ByteChat and combine them with reasoning-heavy models like Claude or GPT for a setup more capable than any single subscription.

Yes — the underlying model is identical. ChatGPT Plus and the OpenAI API use the same GPT model. Claude.ai and the Anthropic API use the same Claude model. The same holds for Gemini, Grok, and every other provider. Consumer subscriptions add extras like a built-in web browser or image generation tools — not a smarter model. ByteChat gives you the same models for a fraction of the cost, and lets you run multiple of them side by side.

ChatGPT Plus: chatgpt.com → Settings → Subscription → Cancel plan. Claude Pro: claude.ai → Settings → Plans → Manage subscription. Gemini Advanced: one.google.com → Manage subscriptions. Perplexity Pro: perplexity.ai → Settings → Subscription → Cancel. Important: cancel before your renewal date. If you subscribed through the iOS or Android app, cancel via Apple ID / Google Play subscriptions instead — not through the website.

ByteChat is a BYOK (Bring Your Own Key) multi-AI chatroom — meaning you connect your own API keys from providers like OpenAI, Anthropic, Google, and more, with no ByteChat markup. Add multiple AI bots to one room and every message broadcasts to all of them by default. Use @BotName to address one specifically. Watch all responses stream in real time. Learn more about BYOK →

BYOK stands for Bring Your Own Key. Instead of paying ByteChat to access AI models, you connect directly to AI providers using your own API keys. Your keys are stored locally in your browser — they never touch our servers. You pay providers directly at their published rates, with no ByteChat markup.

You can try ByteChat with a free account — no API key and no credit card required. Sign up takes under 30 seconds. The Free plan gives you 20,000 characters of conversation across three pre-configured demo bots. When you're ready, add your own API keys — that's free too — to use every provider with unlimited usage.

Each AI provider has their own key page:

• Anthropic (Claude) — console.anthropic.com
• OpenAI (ChatGPT) — platform.openai.com
• Google (Gemini) — aistudio.google.com
• DeepSeek — platform.deepseek.com
• Grok (xAI) — console.x.ai
• Perplexity — perplexity.ai

Most providers give free credits to get started. OpenAI and Anthropic typically give $5–18 of free credits — enough for hundreds of conversations.

See every provider's key dashboard on our How to get an API key page. When you paste a key into ByteChat, it auto-detects the provider for you — no need to pick one.

API usage costs vary by provider and model, but is typically much cheaper than subscription plans for moderate users. For example, GPT-4o-mini costs about $0.15 per million input tokens — a typical conversation message is around 200 tokens, meaning $1 gets you roughly 3,000 message exchanges. Claude Haiku is similarly priced. You only pay for what you use, with no monthly minimum.

In a multi-bot room, your message goes to all active bots by default — just type and send. To address a specific bot only, type @BotName anywhere in your message: @Claude explain quantum computing. You can still target multiple bots explicitly: @Claude @Gpt which is better for coding? — the rest won't reply. With a single bot in the room, messages route automatically either way.

Yes — this is ByteChat's core feature. Add multiple bots to your room and every message goes to all of them simultaneously by default — no @mention required. Each bot streams its answer into its own bubble. Long responses collapse to keep the conversation readable, and you can expand any response with a tap. Use @BotName when you want only one bot to reply.

Tap any bot bubble to reveal the action bar, then tap Reply. A quote strip appears in the input showing the first few words of that bot's response. Your reply is sent to that specific bot with the context of what it said — the bot receives its own prior message as reference, not the entire multi-bot conversation.

Yes — two ways. Right-click (desktop) or long-press (mobile) on any bot chip above the input to open a live model picker for that bot. Alternatively, tap + bots → tap edit next to any bot → select a new model → save. Either way a notification pill appears in the thread marking the model change.

Tap the ■ Stop button that appears below the input while bots are responding. This immediately aborts all active streams — stopping token generation and spend at the provider side.

Your API keys are stored exclusively in your browser's localStorage — on your device only. They are never sent to ByteChat's servers, never stored in any database, and never logged. The only time a key leaves your device is when it's passed directly to the AI provider's API endpoint in the request header, via our stateless proxy.

No. Your conversation messages are passed through our Cloudflare Worker proxy to your chosen AI provider and back. The Worker does not log, store, or process message content — it only relays the stream. Conversations are stored in your browser's localStorage and never leave your device unless you're sending them to an AI provider.

Clearing browser data (localStorage) will remove locally-cached conversations and API keys from that device. If you're signed in, your conversations sync to the cloud and will reappear after you sign back in. Your API keys themselves remain valid with the provider — you'll need to re-enter them after clearing browser data. If you're not signed in, locally-stored conversations are lost permanently on a browser clear.

We recommend using ByteChat only on your personal devices. API keys stored in localStorage are accessible to anyone with physical access to your browser on that device. Always sign out of the browser session when using a shared computer, and consider using a browser profile dedicated to ByteChat.

The Free plan includes: bring-your-own-key for all supported AI providers, unlimited bots per chatroom, 3 free demo bots (no key needed to try), Chain Mode (bots reply in sequence, each building on the last), unlimited saved conversations, 30-day conversation search, cloud sync (access your conversations from any device), Global Memory / Profile Context, Prompt Library, the Subscription Savings Tracker, file attachments (3 files/msg · text, images & PDF), export as Markdown/HTML/PDF, per-bot system prompts, and dark/light theme. Upgrade to Pro ($4.99/mo) for conversation branching, Elevated Thinking, voice-to-multi-bot, live code execution, room templates, full history search, and higher file limits.

Free already gives you the essentials: all supported AI providers via your own key, unlimited bots, Chain Mode, cloud sync, Global Memory, Prompt Library, the Savings Tracker, 30-day search, unlimited conversations, file attachments (text, images & PDF — 3 files/msg), and full export (Markdown, HTML & PDF). Pro ($4.99/mo) adds the power workflows — Consensus / Judge mode (every bot answers, then a judge gives one verdict with where they agree, disagree, and a confidence score), Smart model routing (each question auto-goes to the cheapest capable model, and shows what you saved), Elevated Thinking (a 4-stage protocol where AIs frame independently, pool & de-identify ideas, pressure-test, then synthesise), Conversation Branching (fork any chat without losing the original), a full API Cost Analytics dashboard, voice-to-multi-bot, live code execution, room templates, full history search, unlimited Prompt Library, higher file limits (10/msg), and early access to new features.

Yes. Cancel anytime from your Account Settings → Manage Billing. You'll retain access until the end of your current billing period. No questions asked, no cancellation fees.

We offer a full refund within 7 days of your first payment if ByteChat doesn't meet your expectations. Contact us at [email protected]. After 7 days, we handle refund requests on a case-by-case basis.

Yes — these are two separate bills. ByteChat is free to use with your own keys; the optional Pro plan is a flat $4.99/mo. Your AI provider (OpenAI, Anthropic, etc.) charges separately for API usage at their published per-token rates. ByteChat adds zero markup to provider costs.

ByteChat supports a growing list of providers, including Anthropic (Claude), OpenAI (GPT), Google (Gemini), DeepSeek, Grok by xAI, Perplexity AI, Groq (Llama), Kimi by Moonshot AI, and Qwen by Alibaba. More providers are added regularly — request one at [email protected].

Yes. ByteChat is fully optimised for mobile browsers — Safari on iOS and Chrome on Android. The sidebar collapses into a drawer, the input prevents iOS zoom, and all interactions are touch-friendly. Native iOS and Android apps are in development.

Yes. The Free plan supports 3 files per message — text files (TXT, Markdown, CSV, JSON, code files, and more), image attachments (JPEG, PNG, GIF, WebP) with vision-capable models (Claude, GPT-4o, Gemini), and PDF attachments (text extracted and sent as context). Pro raises the limit to 10 files per message. File content is sent directly to your AI provider — ByteChat never stores your files.

Cross-device sync is included free for every signed-in user. Your conversations and settings sync via Clerk's cloud-backed storage and are available on any device where you sign into your ByteChat account. If you're not signed in, conversations are stored locally in that browser only — create a free account to unlock sync.

Global Memory is a block of text you write about yourself — your role, expertise, preferences, or any context you always want every bot to know. It's automatically injected at the start of every conversation so you never have to repeat yourself. Set it in Settings → Profile. Included free on every plan.

Yes — this is a Pro feature called Room Templates. Once you have your bots, system prompts, and room context set up the way you like, tap the ⊞ template button in the chat header and choose "Save current room as template". Give it a name (e.g. "Research Team" or "Code Review") and it saves the full setup. Load it any time to restore that room in one tap. Great for switching between different AI workflows quickly.

The Prompt Library lets you save prompts you use repeatedly and insert them in one tap. Open it via the 📚 button in the input area. You can drag to reorder prompts and delete ones you no longer need. It's free for everyone; Pro users get unlimited storage.

Bot Chaining is a Pro feature that makes your bots reply in sequence instead of all at once. Each bot reads the previous bots' answers, so a single message can run a pipeline — for example one bot drafts, the next critiques, and a third rewrites. With 2 or more bots active, tap the chain icon next to the mic to turn chain mode on — a Chain order panel opens where you arrange who replies first, second, third (each bot shows its role) using up/down arrows. Send your message and the bots respond one after another in your chosen order, each building on the last. Tap the chain icon again anytime to reorder, or turn chain mode off from inside that panel.

Elevated Thinking is a Pro feature that runs your AIs through a structured 4-stage reasoning protocol from a single question — designed to produce genuinely combined thinking, not just parallel answers. Stage 1 — Independent framings: each AI frames the problem on its own, without seeing the others, so you don't lose good ideas to copying. Stage 2 — Pooled & de-identified: all the ideas are merged into one anonymous list, so they're judged on merit, not authorship. Stage 3 — Lateral pressure: each AI works across the whole pool — merging overlaps, exposing hidden assumptions, and openly revising. Stage 4 — Synthesis: each AI writes one clean summary that keeps the real disagreements instead of forcing false consensus. Stage 5 is you — read the syntheses and pick or braid them forward. With 2 or more bots active, tap the layers icon next to the chain icon to turn it on. Note it runs roughly 4× the API calls of a normal message.

Conversation Branching is a Pro feature that lets you fork a conversation at any message. Tap any message (yours or a bot's) to reveal its action bar, then tap "Branch" — ByteChat creates a new conversation containing everything up to that point. The original conversation stays untouched, so you can explore an alternative direction without losing your thread. Branches appear in your sidebar with a branch tag, and the parent conversation shows how many branches grew from it.

Yes. Pro extends the free Savings Tracker into a full API Cost Analytics dashboard: a 12-month spend bar chart, up to 11 months of history (the free plan keeps a recent summary), a monthly budget you can set with a progress bar and over-budget alert, and one-tap CSV export of all your usage. Open the Usage & Savings panel in settings to see it.