Which AI assistant should you use?

There is no single best assistant, because the six below differ mainly in price, bundled apps and platforms. The short answer by need, using each vendor's own pricing pages as read in October 2026:

• Lowest-priced paid tier: Google AI Plus at $4.99 a month, then ChatGPT Go at $8, which may include ads.
• Already paying for Microsoft 365: Microsoft Copilot, whose paid features come with Microsoft 365 plans from $9.99 a month.
• Answers with sources: Perplexity, which puts inline citations on every answer.
• A Linux desktop app: Claude lists one, in beta. None of the other five lists a Linux app.
• Free plan with a long feature list: Google Gemini or Claude, described below.
• Chat plus a coding agent in one subscription: Le Chat, now Mistral Vibe, at $14.99 a month.

Prices are US figures and change often. Check the vendor's page before paying.

ChatGPT

The Free plan lists unlimited text chats, subject to abuse guardrails, with limited uploads, image creation, voice chats, deep research and memory. OpenAI does not publish numbers for these limits. Paid individual plans are Go at $8 a month, Plus at $20 and Pro from $100. Go may include ads.

ChatGPT runs on the web, iOS and Android, with Mac and Windows downloads and a Chrome extension. Business and Enterprise prices are not shown on the pricing page.

Claude

The Free plan lists web search, file creation, code execution, memory, connectors, Artifacts, voice mode and up to 5 projects. Pro costs $17 a month on an annual plan or $20 billed monthly, and adds Claude Code, Research, Claude in Chrome and unlimited projects. Max starts at $100 a month.

Usage limits apply to every plan and reset on a rolling five-hour window, with weekly limits on paid plans. Message counts are not published, and image generation does not appear in the plan table. Apps exist for web, Mac, Windows, iOS and Android, with Linux in beta.

Google Gemini

The Free plan with a Google Account lists image generation and editing, Deep Research, Gemini Live, Canvas, Gems and 15 GB of storage shared across Gmail, Drive and Photos. Google AI Plus is $4.99 a month with 2x the Free usage limits and 400 GB of storage. AI Pro is $19.99 with 4x the limits and 5 TB. AI Ultra starts at $99.99.

Limits are compute-based and refresh every 5 hours until a weekly limit is reached. Gemini runs on the web, a Windows and macOS desktop app, and mobile.

Microsoft Copilot

Copilot is free with a signed-in Microsoft account, with some features under usage limits that Microsoft does not quantify. There is no Copilot-only paid plan. More usage comes through Microsoft 365: Personal at $9.99 a month, Family at $12.99 for up to 6 people, Premium at $19.99 and Pro at $99.99.

Personal includes Copilot in the desktop Office apps. On Family, the Copilot features are for the subscription owner only. Apps exist for web, Windows, Mac, iOS and Android.

Perplexity

Perplexity combines web search with models from several providers and shows inline citations. The Free plan lists practically unlimited basic searches, 3 Pro Searches a day and 1 Research query a month, with no manual model choice and no image generation.

Pro is $17 a month billed annually and adds Deep Research and model selection; a monthly-billing price was not confirmed. Max is $167 a month billed annually. It runs on the web, iOS and Android.

Le Chat (now Mistral Vibe)

Mistral renamed Le Chat to Vibe and says conversations, settings and plans carried over. The Free plan lists limited messages, limited web searches, limited coding sessions, basic image generation and editing, 100+ connectors and up to 5 scheduled tasks.

Pro is $14.99 a month excluding taxes, or $5.99 for verified students. Its limits are given as multiples of Free, such as up to 6x the messages, rather than absolute numbers. Vibe for code runs in a CLI, in VS Code, JetBrains and Zed, or on the web. The chat app is on web, iOS and Android.

If you want to keep data on your own computer

Hosted assistants process your prompts on the vendor's servers. If that is a problem, run open models locally instead. Ollama is an MIT-licensed model runner for macOS 14 or newer and Windows 10 22H2 or newer; it works mainly from the command line, and the pricing page says running models on your own hardware is always unlimited.

Jan is a free chat app under Apache-2.0 for macOS, Windows and Linux. On Windows and Linux its docs list 8GB of RAM as a minimum and 6GB of VRAM for GPU use. Jan can also connect to cloud providers with your own keys, and prompts sent that way leave your computer. This guide does not compare local models with the hosted assistants above.