Version 2.1 on iPhone and Android. Analyze mode on the web.
Ten AIs in one app. They keep each other honest.
Where Ideas Converge. Where Understanding Emerges.
ChatGPT, Claude, Gemini, and seven more, with your own API keys. Chat with several at once to guard against hallucinations, compare two answers side by side, watch AIs debate, and create images, video, and audio. On the web, hand a question and a dataset to an AI that runs the analysis while a second AI checks its work.
20 free sessions on the web with your own keys. Demo Mode on mobile needs no account at all.
Works with
- Claude
- ChatGPT
- Gemini
- Perplexity
- Mistral
- Cohere
- DeepSeek
- Grok
- Kimi
- GLM
- Runway (video)
- ElevenLabs (audio)
01New on the web
Analyze: an AI that shows its work.
Give it a question and your data. One AI writes and runs the code, a second AI from a different provider critiques the result, and you get one report with every figure traced to its source. Agentic AI, with a human at the gate.
- 01You set the table
- Upload files (CSV, Excel, JSON, PDF, images, up to 100 MB each), pick from 38 built-in data connectors, choose a report type or leave it on Auto, and ask in plain language.
- 02The Operator works
- One AI runs a genuine tool-use loop: it writes and runs Python in your browser, queries SQL, searches the web, and fetches data, then continues from its own results until the job is done. Your Stop button bounds it.
- 03The Reviewer pushes back
- A second AI from a different provider reads the result and returns up to five prioritized critiques. It cannot run tools or touch the report. You choose which items go back to the Operator.
- 04One report, every format
- The Operator authors a single structured report. Publish it as PDF, PowerPoint, Word, or an interactive page, plus an Excel workbook of the data behind it. Every sourced figure carries its provenance.
- Your files stay in your browser
- Uploaded data mounts into an in-browser Python runtime. The model sees schemas, short previews, and what the code prints, not the files.
- Provenance is captured, not curated
- Every fetch records the request, the time, and a response fingerprint. They flow into inline markers, a Sources panel, and an appendix in every export.
- Real data only
- The Operator is instructed never to invent, sample, or guess values, and never to fall back silently to training data. If it cannot get the data, it says so.
Analyze needs a provider with function calling: Claude, ChatGPT, Gemini, Grok, Mistral, Cohere, Kimi, or GLM. Web only for now.
02Version 2.1 on mobile
Composer-first, everywhere.
Chat, Compare, Debate, and Create all begin at a single prompt bar. Tap to pick your AIs, attach what you need, and go. No setup screens between you and the conversation.
- Pick providers and models with a tap. Each pill shows what it will run and what it costs.
- Attach images and documents before the first message.
- Sending starts the session. There is no separate Start button.
- A live status line while responses stream in, and smoother keyboard handling on every screen.
03The modes
Five rooms, one prompt bar.
Each mode has its own room and its own name in the app. All of them start the same way.
01Multi-AI chat
The Forum
Pick up to three AIs and get answers from all of them in one conversation, so they can check each other in real time.
02Compare
The Lens
Ask one question, get two answers side by side, then continue with the one you preferred.
03Debate
The Arena
Pick a motion, a format, and two to four debaters, then watch them argue live with scoring, audience voting, and leaderboards.
04Create
The Studio
Generate images, video, and audio from one composer, across multiple providers, with a gallery that keeps every creation a tap away.
05History
The Archives
Every chat, comparison, debate, and analysis is saved and can be picked up where you left off.
04On a tablet
Room to think.
On iPad and Android tablets the tabs become a sidebar and comparisons get the width they deserve.
05What's inside
The details that add up.
- Web search with citations
- Claude, ChatGPT, Gemini, Grok, and Perplexity search the live web automatically and answer with numbered, cited sources. No toggle to flip.
- Attach before you ask
- Add images and documents (PDF, Word, Excel, PowerPoint, CSV, and more) to your very first message so the AIs have context from the start.
- Create Studio
- Images from OpenAI, Gemini, and Grok, with up to three models at once and photo-based refinement. Video clips through Runway. Voiceovers and sound effects through ElevenLabs.
- Debate formats and voices
- Oxford, Lincoln-Douglas, and Policy formats with Friendly, Neutral, or Hostile intensity. Give each debater an ElevenLabs voice and turn the debate into a podcast.
- Eight AI personalities
- From Bestie, the empathetic co-pilot, to Devlin, the respectful counter-argument artist. A style layer that shapes tone without changing the model.
- Expert Mode
- Set default models per provider and tune model-aware parameters like temperature and max tokens, globally or per session, with live pricing so you always know the cost.
06Pricing
Your keys. Provider prices. One $5.99 app.
Symposium AI connects to each provider with your own API key. You pay OpenAI, Anthropic, Google, and the others exactly what they charge, with no markup and no usage caps, and $5.99 a month for the app. Pick a small, fast model for small jobs and a frontier model when it matters, with the price shown before you send.
Try it free
$0
Before the trial starts
- Web: 20 free sessions with your own keys, 5 each for chat, compare, debate, and analyze.
- Mobile: Demo Mode plays recorded chats, comparisons, and debates. No account or keys needed.
Premium
$5.99 a month
7-day free trial, all platforms
- Chat, Compare, Debate, Create, and Analyze with every supported provider
- Web search with cited sources
- Image, video, and audio generation
- Attachments, Expert Mode, and all eight personalities
- Report exports: PDF, PowerPoint, Word, Excel, interactive page
- Web, iPhone, and Android
On mobile, also $49.99 a year or $129.99 once for lifetime access. Cancel anytime.
API usage
Pay as you go
Billed by each provider
- Pay OpenAI, Anthropic, Google, and the others directly, only for what you use.
- Web: keys are stored AES-256 encrypted on the server and never exposed to the browser.
- Mobile: keys stay on your device.
- Live model pricing in the app, so you always know the cost.
If you run frontier models all day, your API bill will look like a subscription. You will still have ten providers in one place and a second opinion on every answer. Live sessions require your own provider API keys plus free sessions, an active trial, or a subscription.
Start in a browser tab, or on your phone.
Sign in, add a key, and start typing. Or explore Demo Mode on mobile with no account.