Imgica AI
Opening ai chat.0.0s
The chat studio
One AI chat window,
every frontier line.
Claude, GPT, Gemini, GLM, MiniMax, Qwen and Nova, answering the same question from one window. Switch the model mid-thread, and web search, artifacts and personas ride along either way.
Free tokens at signupNo card requiredPay once, from $19Commercial license included
imgica.com / studio / chatReal replies
Build a pricing table with three tiers and one highlighted total.
Claude Opus 4.8 · Anthropic
I'll create a responsive three-column pricing table with clean card styling, and make the middle tier stand out with a highlighted border, accent color, and a "recommended" badge to draw attention.
Artifactsindex.html rendering live beside the chat
Every reply above is real: this exact question went to each model on its nameplate, and the answers print exactly as they came back.
01 · The roster
Seven labs, one AI chat.
Every line draws on the same pool of tokens. Ask once, then pick who answers, or switch mid-thread without losing the conversation.
01AnthropicClaude Opus 4.8 · Opus 4.7 · Opus 4.6 · Claude Fable 5Extended thinking on the questions that deserve it.4 on the line
02OpenAIGPT 5.2 · GPT 5.5 · ChatGPT ProFast iteration with tools built in.3 on the line
03GoogleGemini 3.1 Pro · Gemini 3 ProLong context, tuned for research.2 on the line
04ZhipuGLM 5.2 · GLM 5Lean and quick on code.2 on the line
05MiniMaxMiniMax M2.5Fast first drafts, low friction.1 on the line
06VisionAlibabaQwen3 VLReads screenshots, photos and diagrams.1 on the line
07AmazonNova 2Efficient on plain, everyday asks.1 on the line
New lines join the roster the week they ship. No plan change, no re-onboarding.
Start on any line02 · Openers
Bring these, see what comes back.
Real openers, printed as typed. Each slip names the line it suits, in the roster's own words. No replies printed here; the switchboard above already answers for real.
Opener 01Amazon
Rewrite this email so it sounds like a human wrote it.
Hand it toNova 2
efficient on plain, everyday asks
Opener 02Zhipu
Explain this error message like I am five.
Hand it toGLM 5.2
lean and quick on code
Opener 03OpenAI
Turn these bullet points into a one-page brief.
Hand it toGPT 5.2
fast iteration with tools built in
Opener 04MiniMax
Draft the launch tweet in three registers.
Hand it toMiniMax M2.5
fast first drafts, low friction
Opener 05Anthropic
Argue both sides of this decision, then tell me which side you would take.
Hand it toClaude Opus 4.8
extended thinking, on a question that deserves it
Opener 06Google
Read the three articles I pasted and pull out where they disagree.
Hand it toGemini 3.1 Pro
long context, tuned for research
Type one as written, or hand it to a different line and compare.
Take a slip to the studio03 · Artifacts
Code that renders while you watch.
Ask for a landing page, a small tool or a chart, and the artifact opens beside the thread as a live preview instead of a wall of code. Change your mind in plain language and the preview keeps up.
- Renders
- Code and full pages, live
- On a fault
- Caught and repaired automatically
- Where it sits
- Beside the thread, not below it
imgica.com / studio / chat / artifacts
“Build the pricing table, three tiers”Preview
index.htmlLiveWrite→Render→Fault→Repair→Render
04 · Personas
Personas hold the memory, Vani holds the room.
A persona is a standing brief the chat keeps on file, so a new conversation never starts from zero.
A standing brief
Create a persona once and every new conversation opens already briefed: who you are, what you are working on, how you like the answer shaped.
Vani speaks the room
Vani personas are built into the chat and hold a conversation in about 30 Indian languages, from Hindi and Tamil to Bhojpuri. No separate app, just a persona you open.
Persona register
Custom personasmemory of your context
Vani personasabout 30 Indian languages
Model choiceany line on the roster
Where it livesinside every line.
05 · Search
Web search and deep research.
Two depths, chosen per question rather than baked into a plan.
Mode A
Live web search
- Depth
- one pass
- Returns
- a current answer
The chat reads the live web before it answers, so the reply can lean on what is happening right now instead of stopping at training data.
Mode B
Deep research
- Depth
- extended run
- Returns
- organized findings
For the questions one search cannot settle. The chat runs a longer pass across sources and reports back with the findings organized.
chosen per question, not per plan
One chat workspace with the frontier models from Anthropic, OpenAI, Google, Zhipu, MiniMax, Alibaba and Amazon behind a single login. You pick which model answers, and the conversation, artifacts, personas and web search all live in the same place.
Claude Opus 4.8, 4.7 and 4.6 plus Claude Fable 5 from Anthropic; GPT 5.2, GPT 5.5 and ChatGPT Pro from OpenAI; Gemini 3.1 Pro and Gemini 3 Pro from Google; GLM 5.2 and GLM 5; MiniMax M2.5; Qwen3 VL with vision; and Amazon Nova 2.
Yes. Live web search is built in, so answers can draw on what is happening right now. For bigger questions there is deep research, which runs a longer investigation and reports back organized findings.
On models that support image input, yes. Drop in a screenshot, photo or diagram and ask about it. Qwen3 VL is built for vision, and several other models on the roster read images too.
Through personas, yes. You can create custom personas that keep memory of facts about you, so every new conversation starts already briefed. The built-in Vani personas also chat in about 30 Indian languages, from Hindi and Tamil to Bhojpuri.
You get free tokens at signup, no card required, and every model on the roster is available from day one. When you outgrow the free tokens, plans are one time payments rather than subscriptions.
07 · Begin
Pick up the line.
Start creating free
Free tokens at signup · no card required · everything you make is yours