Why is one service unlimited and another metered?

Every reply costs computing, and the cost grows with the size of the model and the time it spends thinking. Free, unlimited chats run on small models. The large reasoning models are rationed on every service, at every price; the difference is whether you can see the ration. Ethocana shows it as a meter with reset times.

Which model should I use?

The smallest one that does the job. Use a lightweight model for rewriting, summaries and quick lookups, a mid-size model for most work with files and several steps, and the largest only for long documents and hard reasoning. A useful habit: have the small model write a full, careful prompt, then send that prompt to the large one.

Are my conversations used to train AI?

It depends on the service and the plan. On consumer plans, ChatGPT and Claude may use your chats to train future models unless you opt out in settings; business plans generally do not by default. Ethocana does not use conversations to train models, and neither does the provider of its models.

Is a bigger context window always better?

No. A larger window holds more, but the model pays less attention to any one part of it, and every reply rereads the whole thing, so long chats use more allowance. Attach the documents that matter, and start a new chat when the subject changes.

Can a chat AI do arithmetic reliably?

On its own, a language model estimates numbers the way a person does mental arithmetic, and it can slip. Services that run code, or that carry a calculation engine as Ethocana does, compute the figure exactly. For any number you will act on, ask the AI to show its working.

Chat AI Explained: ChatGPT, Claude, Gemini and More Compared

Every Category.

The same 42 categories as the map, by layer. Choose one to open its card.

Conversation

the chat itself

Chats Models Effort Instructions Sharing

Reading

what it takes in

Documents Images Video Audio files

Research

the live web

Web search Page reading Deep research

Making

what it hands back

Writing File creation Slides Canvas Calculation Code execution

Media & Voice

pictures, sound, motion

Image generation Image editing Video generation Music & audio Voice Camera & screen

Memory & Context

what it keeps

Memory Past chats Projects Context Custom assistants

Agents & Automation

work done for you

Helpers Task lists Schedules Browser agents Connectors Approvals

Plans & Platforms

where, and how much

Plans Free plan Team plans Usage limits Privacy Desktop apps Mobile apps

How a Chat AI Works.

Three parts sit behind every reply: the model, the workspace around it, and the meter.

The model writes.

A large language model is a very large file of numbers, trained to continue text. Before each reply it reads the whole conversation, then writes the answer a few characters at a time. It knows what it learned in training and nothing newer, and it does arithmetic by estimate.

The workspace gives it reach.

Everything else belongs to the workspace: the search that brings in today’s pages, the reader that opens your PDF, the engine that computes a figure exactly, the memory list and the files it hands back. Two services on similar models can feel very different, because the workspace decides what the model can reach.

The meter rations it.

Each reply costs computing in proportion to the text read and written, and the thinking in between. Plans ration that cost by time windows. Large models, high effort, helpers, research and long chats spend it fastest.

The model is the engine; the workspace is the rest of the car.

The Services, Plan by Plan.

Choose a service to see what it does best and what each plan costs.

Ethocana^©Autocomputing

An AI workspace for thinking, writing, researching and building. You choose the model, the effort and the tools for every message, and a meter shows exactly how much of your allowance is left and when it refills.

What it does best

Model, effort and three tools set separately, per message
Exact figures from a built-in calculation and finance engine
A visible meter with reset times, and a Usage chart
Approval cards before any change to GitHub or web hosting
Scheduled tasks that email the result

Plan	Price	Models	Usage
Ethocana	$16 a month, or $160 a year	Simplex, Orthoplex, Dodecaplex	Five-hour window, weekly window and monthly cap, each shown with its reset time

ChatGPTOpenAI

The largest chat AI service, with the widest set of tools: images, voice with video, deep research, a built-in browser, Codex for code, and ChatGPT Work for longer jobs on the desktop.

What it does best

Unlimited everyday text chats on the free plan, within abuse guardrails
Image creation on every plan
Voice with video on Go and above
Codex and ChatGPT Work for long, agent-run jobs

Plan	Price	Models	Usage
Free	$0	GPT-6 Luna	Unlimited text chats; limited uploads, images, voice and deep research
Go	About $8 a month	GPT-6 Luna	More messages with tools, uploads, images and voice; may include ads
Plus	About $20 a month	GPT-6 Astra, GPT-6.1 Sol, GPT-6 Sol and others	Expanded; limits apply but are not published
Pro	About $100, $200 or $500 a month	Adds Pro reasoning on GPT-6 Astra	Three usage tiers; unlimited image creation; the most deep research
Business	Per seat, from 2 users	As Plus, with workspace controls	Credits can be bought for more access

ClaudeAnthropic

A chat service known for writing and for code, with Artifacts beside the chat, Research, connectors and, on paid plans, Claude Code and Claude in Chrome.

What it does best

Context windows of up to 1 million tokens
Artifacts, Slides and file creation on every plan
Research and Claude Code on paid plans
Usage bars, and paid usage credits to continue past a limit

Plan	Price	Models	Usage
Free	$0	Sonnet, Haiku	Limited, every five hours and every week
Pro	$20 a month, or $200 a year	Opus, Sonnet, Haiku; Fable on usage credits	The baseline: one Pro of usage per five hours and per week
Max	$100 a month	Opus, Sonnet, Haiku, Fable	5 times Pro, every five hours and every week
Ultra	$200 a month	Opus, Sonnet, Haiku, Fable	10 times Pro each week; no five-hour limit for verified accounts
Team	$25 or $125 per seat a month	Opus, Sonnet, Haiku	Premium seats carry 5 times the usage of the $25 seat

GeminiGoogle

Google’s AI, joined to Search, Gmail, Docs, YouTube and Android, with the broadest media tools: Nano Banana images, Veo video, music, and live camera conversations.

What it does best

Images, music and live camera on the free plan
Video generation on paid plans
Google Search, YouTube and Workspace in one assistant
Cloud storage bundled with every paid plan

Plan	Price	Models	Usage
Free	$0	Flash-Lite	Refreshes every five hours until a weekly limit; 32,000-token context
AI Plus	$4.99 a month	Flash	2 times the free allowance; 128,000-token context; 400 GB storage
AI Pro	$19.99 a month	Pro and Flash	4 times the free allowance; 1 million-token context; 5 TB storage
AI Ultra	$99.99 or $199.99 a month	Highest access to every model	5 or 20 times AI Pro; from 20 TB storage

GrokxAI

xAI’s assistant, joined to X for live posts and news, with Grok Imagine for pictures and video and a multi-agent Heavy mode.

What it does best

Live information from X
Image and video generation in one allowance
Grok Heavy, several agents working one problem

Plan	Price	Models	Usage
Free	$0	Grok	Limited
SuperGrok	About $30 a month	Grok, with reasoning modes	One weekly allowance shared by chat, Imagine, voice and building
SuperGrok Heavy	About $300 a month	Adds Grok Heavy	The largest weekly allowance

CopilotMicrosoft

Microsoft’s assistant, built into Windows, Edge and the Office apps, with voice, screen reading and actions on the PC itself.

What it does best

Built into Windows 11, with a wake word
Copilot Vision reads what is on screen
Word, Excel, PowerPoint and Outlook with Microsoft 365

Plan	Price	Models	Usage
Free	$0	Microsoft’s current models	Limits not published
Microsoft 365 Premium	$19.99 a month, or $199.99 a year	Adds Researcher and Analyst	Higher limits, described as extensive; includes the Office apps

PerplexityPerplexity

An answer engine: every reply is a researched answer with sources. Paid plans pick from several makers’ models, and the Comet browser carries the assistant onto any page.

What it does best

Citations on every answer
Models from several makers in one picker
The Comet browser, free to everyone

Plan	Price	Models	Usage
Free	$0	Perplexity’s default model	Limited Pro searches and research
Pro	About $20 a month	Choice of models from several makers	More Pro searches, research and uploads
Max	About $200 a month	Every model offered	The highest research and browser-agent limits

Le ChatMistral

Mistral’s assistant from France, with fast answers, web search, images, a canvas and saved agents.

What it does best

Fast replies
A European company and data location
Saved agents and a canvas

Plan	Price	Models	Usage
Free	$0	Mistral models	Limited
Pro	About $14.99 a month	Mistral’s largest models	Higher limits
Team and Enterprise	Per seat	Mistral models	Workspace controls

Z.aiZhipu

Zhipu’s GLM models in a free chat, with modes that build a full-stack app, a website, a slide deck, a piece of writing or a data analysis from one prompt.

What it does best

Free chat on GLM-5.3-Flash
One-prompt websites, apps and slide decks
A 1 million-token context on GLM-5.3

Plan	Price	Models	Usage
Chat	$0	GLM-5.3-Flash; GLM-5.3	Not published
GLM Coding Plan	From $18 a month	GLM-5.3, GLM-5.3-Flash, for ZCode and coding tools	Credits per five hours and per week: Lite 2,000 and 10,000, Pro 12,000 and 60,000, Max 28,000 and 140,000; half rate off-peak

DeepSeekDeepSeek

A free chat on DeepSeek’s own models, with a reasoning switch and web search.

What it does best

Free, with no paid chat plan
Strong mathematics and code for the price

Plan	Price	Models	Usage
Chat	$0	DeepSeek V4	Not published

US list prices on November 7, 2026. “About” marks a price that the service’s own plan page did not print; it comes from independent price trackers that agree with one another. Usage is described in each service’s own terms.

What Each Plan Costs.

Every plan of every service, on one scale.

Monthly list prices in US dollars. Paid plans on a logarithmic scale, so each step to the right is a multiple, not an addition. Annual discounts are left out.

Capabilities, Side by Side.

Ethocana and 9 other services, capability by capability. Choose a capability to open its card.

On every planPaid plansNot advertised

Capability	Ethocana	ChatGPT	Claude	Gemini	Grok	Copilot	Perplexity	Le Chat	Z.ai	DeepSeek
Conversation
Free plan	Not advertised	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan
Choice of models	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	Paid plans	On every plan	On every plan	On every plan
Effort or thinking control	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan
Share links	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan
Reading
Documents and PDFs	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan
Images and screenshots	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan
Uploaded video	On every plan	Not advertised	Not advertised	On every plan	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised
Audio files	Not advertised	Not advertised	Not advertised	On every plan	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised
Research
Web search with sources	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan
Deep research reports	On every plan	On every plan	Paid plans	On every plan	On every plan	On every plan	On every plan	On every plan	Not advertised	Not advertised
Making
Word, PDF and Excel files	On every plan	On every plan	On every plan	On every plan	Not advertised	On every plan	Paid plans	Not advertised	On every plan	Not advertised
Slides	Not advertised	On every plan	On every plan	On every plan	Not advertised	On every plan	Paid plans	Not advertised	On every plan	Not advertised
Canvas and live previews	Not advertised	On every plan	On every plan	On every plan	Not advertised	Not advertised	Not advertised	On every plan	On every plan	Not advertised
Calculation and finance engine	On every plan	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised
Code execution	Not advertised	On every plan	On every plan	On every plan	Not advertised	Not advertised	Not advertised	On every plan	On every plan	Not advertised
Media and voice
Image generation	Not advertised	On every plan	Not advertised	On every plan	On every plan	On every plan	Paid plans	On every plan	Not advertised	Not advertised
Image editing	Not advertised	On every plan	Not advertised	On every plan	On every plan	On every plan	Not advertised	Not advertised	Not advertised	Not advertised
Video generation	Not advertised	Not advertised	Not advertised	Paid plans	Paid plans	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised
Music generation	Not advertised	Not advertised	Not advertised	On every plan	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised
Voice conversation	Not advertised	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	Not advertised	Not advertised
Live camera and screen share	Not advertised	Paid plans	Not advertised	On every plan	On every plan	On every plan	Not advertised	Not advertised	Not advertised	Not advertised
Memory and context
Memory you can edit	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	Not advertised	Not advertised
Past-chat search	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	Not advertised	Not advertised	Not advertised	Not advertised
Projects or folders	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	Not advertised	Not advertised
Custom assistants	Not advertised	Paid plans	Not advertised	On every plan	Not advertised	Not advertised	Not advertised	On every plan	Not advertised	Not advertised
Agents and automation
Helper AIs, set by you	On every plan	Not advertised	Paid plans	Not advertised	Paid plans	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised
Scheduled tasks	On every plan	Paid plans	Paid plans	Paid plans	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised
Browser or computer agent	Not advertised	On every plan	Paid plans	Paid plans	Not advertised	On every plan	On every plan	Not advertised	Not advertised	Not advertised
Connectors	On every plan	On every plan	On every plan	On every plan	Not advertised	On every plan	On every plan	On every plan	Not advertised	Not advertised
Plans and platforms
Meter with reset times	On every plan	Not advertised	On every plan	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised
Team plans	Not advertised	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	Not advertised	Not advertised
Desktop app	Not advertised	On every plan	On every plan	On every plan	Not advertised	On every plan	On every plan	Not advertised	Not advertised	Not advertised
Mobile app	Not advertised	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	On every plan	Not advertised	On every plan

Checked against each service’s web app and plan pages on November 7, 2026. Features change often; a dash means the service does not advertise the capability, not that it can never do it.

How Limits Are Set.

Every service rations usage. Few publish the ration.

Most paid plans now count usage in two windows: a short one, usually five hours, that stops a single afternoon from using the week, and a weekly one. A few add a monthly cap. Services seldom say how many messages fit in a window, because the answer depends on the model, the effort, the files attached and the length of the chat; they publish multiples instead, such as five times the usage of the plan below.

Service	Short window	Long window	What you see	When it runs out
Ethocana	Five hours	One week, and a monthly cap	A meter for all three windows, each with its reset time, and a Usage chart	Pauses until the window reopens; nothing is charged extra
ChatGPT	Varies by model and tool	Not published	A notice when a limit is reached	Wait for the limit to reset, or move to a higher plan
Claude	Five hours (none on verified Ultra)	One week, plus a separate weekly limit for Fable	Usage bars in Settings	Wait, or continue on paid usage credits at API rates
Gemini	Five hours	One week	A notice when a limit is reached	Wait; limits are multiples of the free allowance
Grok	None	One week, shared by chat, images, video and voice	The weekly allowance	Wait for the week to reset
Z.ai Coding Plan	Five hours	One week	Credits used and left	Wait; off-peak use costs half the credits
Copilot, Perplexity	Daily, for some features	Not published	A notice when a limit is reached	Wait, or move to a higher plan

Controls and Interface.

What you can set before you press send.

	Ethocana	ChatGPT	Claude	Gemini	Grok	Z.ai
Model	Three models; switch mid-chat	Model picker, by plan	Opus, Sonnet, Haiku; Opus on paid plans	Model picker, by plan	Model picker, by plan	GLM-5.3 or GLM-5.3-Flash
Effort	Hare, Fox or Tortoise, per message	Instant or Thinking	Extended thinking on or off	Thinking modes	Reasoning modes	Thinking on or off
Tools	Search, Helpers and Memory, each with levels, per message	A tools menu: search, deep research, images, agent	Search, Research and connector switches	A tools menu	DeepSearch, Imagine	Mode chips: Full-Stack, Writing, Data Insight, PPT, Web Site
Reply length	A Length knob	Through instructions and personalities	Through styles	Through instructions	Through instructions	Through the prompt
Usage shown	Three windows with reset times; Usage chart	At the limit	Usage bars	At the limit	Weekly allowance	Coding Plan credits
Commands	Slash commands for model, effort, memory, sharing and more	Not advertised	Slash commands for skills	Not advertised	Not advertised	Not advertised
Per-reply detail	Developer Pane: context fill, tokens, lookups and saved steps	Not advertised	Not advertised	Not advertised	Not advertised	Not advertised

On the Desktop and the Phone.

What the installed apps add to the web version.

Service	Desktop	Phone	What the installed app adds
Ethocana	Unavailable	Unavailable	Runs in current Chrome, Edge, Firefox and Safari.
ChatGPT	Mac, Windows	iOS, Android	ChatGPT Work for long jobs, Codex, a built-in browser, and work alongside other apps. The separate Atlas browser closed in August 2026 and its browsing moved into the desktop app.
Claude	Mac, Windows	iOS, Android	Chat and Cowork in one app since September 2026, Claude Code, desktop extensions for local tools, and computer use in beta on paid plans.
Gemini	Mac, Windows	iOS, Android	Alt + Space opens Gemini over any window. Spark, an agent that works with files on the computer, is in beta on the Mac for AI Ultra in the US.
Copilot	Built into Windows 11; Mac	iOS, Android	The wake word “Hey Copilot”, Copilot Vision to read the screen, and Copilot Actions to carry out tasks in apps, switched on by the user.
Perplexity	Comet browser, Mac and Windows	iOS, Android	Comet is a whole browser with the assistant built in; it reads and acts on the open page.
Grok	Not advertised	iOS, Android	Grok Imagine and voice on the phone.
Le Chat	Not advertised	iOS, Android	Voice and camera input on the phone.
Z.ai	ZCode, for coding	Not advertised	ZCode is a coding agent with approval modes for edits and commands.
DeepSeek	Not advertised	iOS, Android	The same chat as the web.

Installed apps reach what a web page cannot: a keyboard shortcut over any window, files on the computer, the screen, and the microphone and camera without a browser between.

Five Questions About Chat AI.

The ones we are asked most.

Why is one service unlimited and another metered?
Every reply costs computing, and the cost grows with the size of the model and the time it spends thinking. Free, unlimited chats run on small models. The large reasoning models are rationed on every service, at every price; the difference is whether you can see the ration. Ethocana shows it as a meter with reset times.
Which model should I use?
The smallest one that does the job. Use a lightweight model for rewriting, summaries and quick lookups, a mid-size model for most work with files and several steps, and the largest only for long documents and hard reasoning. A useful habit: have the small model write a full, careful prompt, then send that prompt to the large one.
Are my conversations used to train AI?
It depends on the service and the plan. On consumer plans, ChatGPT and Claude may use your chats to train future models unless you opt out in settings; business plans generally do not by default. Ethocana does not use conversations to train models, and neither does the provider of its models.
Is a bigger context window always better?
No. A larger window holds more, but the model pays less attention to any one part of it, and every reply rereads the whole thing, so long chats use more allowance. Attach the documents that matter, and start a new chat when the subject changes.
Can a chat AI do arithmetic reliably?
On its own, a language model estimates numbers the way a person does mental arithmetic, and it can slip. Services that run code, or that carry a calculation engine as Ethocana does, compute the figure exactly. For any number you will act on, ask the AI to show its working.

Chat AI, Explained.

Every Category.

Conversation

Reading

Research

Making

Media & Voice

Memory & Context

Agents & Automation

Plans & Platforms

How a Chat AI Works.

The model writes.

The workspace gives it reach.

The meter rations it.

The Services, Plan by Plan.

Ethocana©Autocomputing

ChatGPTOpenAI

ClaudeAnthropic

GeminiGoogle

GrokxAI

CopilotMicrosoft

PerplexityPerplexity

Le ChatMistral

Z.aiZhipu

DeepSeekDeepSeek

What Each Plan Costs.

Capabilities, Side by Side.

How Limits Are Set.

Controls and Interface.

On the Desktop and the Phone.

Five Questions About Chat AI.

Why is one service unlimited and another metered?

Which model should I use?

Are my conversations used to train AI?

Is a bigger context window always better?

Can a chat AI do arithmetic reliably?

A Simpler Way to AI.

Chats & Conversations

Model Choice

Effort & Thinking Time

Instructions & Reply Length

Sharing & Export

Documents & PDFs

Images & Screenshots

Video Files

Audio & Voice Notes

Web Search & Sources

Web Page Reading

Deep Research Reports

Writing & Editing

Word, PDF & Excel Files

Slides & Presentations

Canvas & Live Previews

Exact Calculation & Finance

Code Execution

Image Generation

Image Editing

Video Generation

Music & Audio Generation

Voice Conversation

Live Camera & Screen Share

Memory

Past-Chat Search

Projects & Folders

Context Window

Custom Assistants

Helper AIs (Sub-Agents)

Task Lists & Questions

Scheduled Tasks

Browser & Computer Agents

Connectors & Integrations

Approvals Before Changes

Plans & Pricing

Free Plan

Team & Business Plans

Usage Limits & Meters

Privacy & Account Security

Desktop Apps

Mobile Apps

Ethocana^©Autocomputing