ChatGPT: Features, Models, Pricing, and Alternatives

Last researched: September 26, 2026. Pricing and model names change frequently — figures below were cross-checked against OpenAI's own announcement pages and multiple independent trackers current as of that date, and you should verify anything price-sensitive against openai.com and chatgpt.com/pricing before acting on it.

ChatGPT is the conversational AI product made by OpenAI. It was the product that took generative AI mainstream after its November 2022 launch, and it remains the most widely used general-purpose AI assistant, running on OpenAI's GPT family of large language models.

What Is ChatGPT?

ChatGPT is a chat-based interface to OpenAI's language models. You type (or speak) a request, and it responds with text, and — depending on the model and plan — can also search the web, read files and images you upload, write and run code, generate images, and take multi-step actions on your behalf through connected apps. It's available as a website, desktop and mobile apps, a browser extension, and a developer API that other products build on.

Key Capabilities

• Conversational reasoning — GPT-6 Astra (OpenAI's current flagship, released September 2026) and the GPT-5.6 family (Sol, Terra, Luna) that preceded it support an explicit "thinking"/reasoning mode that trades latency for better performance on multi-step problems, alongside a faster "instant" mode for everyday chat.
• Vision and file understanding — upload images, PDFs, spreadsheets, and other documents for ChatGPT to read, summarize, extract data from, or reason about.
• Web search — ChatGPT can search the live web and cite sources for answers about current events or anything past its training data.
• Image generation — built-in image generation (GPT Image) lets you create and edit images from a text prompt directly inside a chat.
• Code execution and Codex — a sandboxed code interpreter for running Python, plus deeper access to OpenAI's Codex coding agent on paid plans for larger software tasks.
• Agentic tasks — "ChatGPT Work" (introduced 2026) and Deep Research let ChatGPT plan and execute multi-step tasks — gathering information across connected apps and files to produce documents, spreadsheets, and slide decks, or to compile a long-form research report with citations.
• Memory — ChatGPT can remember facts and preferences across conversations (on by default, and user-editable/erasable in settings).
• Custom GPTs — build a version of ChatGPT configured with custom instructions, a knowledge base, and specific tools, and share it with others.

Main Features

Use Cases

• Software development — through Codex, ChatGPT can scaffold, refactor, debug, and explain code across most mainstream languages.
• Content creation — drafting, editing, and rewriting articles, marketing copy, emails, and scripts.
• Research — Deep Research compiles multi-source reports with citations for a given question.
• Business — Projects and ChatGPT Work help teams draft documents, spreadsheets, and presentations from connected company data.
• Education — explaining concepts, tutoring through problems step by step, and summarizing reading material.
• Data analysis — uploading a spreadsheet or CSV and asking ChatGPT to clean, analyze, or chart it via its code interpreter.

Not genuinely supported: ChatGPT does not generate video natively inside the product (OpenAI's video model, Sora, is a separate product), and it isn't a dedicated design tool — image generation is conversational, not a layered editing canvas like Photoshop or Figma.

Supported Models / Technologies

OpenAI's model lineup moves quickly. As of September 2026:

• GPT-6 Astra — OpenAI's newest and most capable model, released in a limited preview on September 3, 2026 and rolling out to ChatGPT Plus, Pro, Business, and Enterprise shortly after. OpenAI describes it as a major step up in reasoning, coding, computer use, and research, with an explicit reasoning mode. Context window: roughly 1 million tokens, per independent model trackers (not confirmed on OpenAI's own product page at time of writing).
• GPT-5.6 family (Sol, Terra, Luna) — released July 2026; Sol is the flagship, Terra a balanced mid-tier model, and Luna a low-cost, high-speed model. These remain in use for lower-tier plans and cost-sensitive API use even after GPT-6 Astra's release.
• GPT-5.2 Instant — the model behind the budget "Go" plan.
• Multimodal support: text, images, and documents as input; text and image as direct output (video and long-form audio are handled by separate OpenAI products, not inline in a chat turn).
• Agents/tool use: function calling, structured JSON output, and connected-app tool use are supported via both ChatGPT's built-in integrations and the API's tools/tool_choice parameters.

Because OpenAI revises its model lineup every few months, treat specific model names here as a snapshot — check OpenAI's model release notes for the current default model.

Platforms

• Web (chatgpt.com)
• Windows desktop app
• macOS desktop app
• iOS app
• iPadOS app
• Android app
• API (OpenAI API, used by third-party products)
• Browser extension (ChatGPT's Chrome/Edge extension — OpenAI's standalone "Atlas" browser was discontinued on August 9, 2026 in favor of this extension and the desktop app)

No official Linux desktop app exists as of this writing; Linux users access ChatGPT through the web app.

Integrations

ChatGPT can connect to external apps and data sources — including common productivity and cloud-storage tools — so ChatGPT Work and Deep Research can pull in files and context from them when producing documents or answers. Developers can also build against the OpenAI API to embed GPT models (including function/tool calling) into their own products, and Codex integrates directly with developer environments including the CLI and IDE extensions (e.g. VS Code).

Pricing

(All prices in USD; last checked September 26, 2026. OpenAI's own pricing pages could not be fetched directly while researching this article — figures below are cross-referenced across several independent pricing trackers rather than confirmed on a single official source, so verify current numbers at chatgpt.com/pricing before relying on them.)

Advantages

• Broadest name recognition and largest third-party ecosystem of any AI assistant, meaning the most tutorials, integrations, and community support.
• Fast-moving model lineup — new flagship models ship every few months.
• Wide platform coverage (web, desktop, mobile, API, browser extension).
• Strong all-in-one feature set: chat, search, images, code execution, agentic tasks, and a coding agent in one product.

Limitations

• No native Linux desktop app.
• No inline video generation in the chat product itself (handled by a separate OpenAI product, Sora).
• Free and Go tiers do not include the flagship reasoning model.
• Model names and pricing tiers change often enough that published comparisons (including this one) can go stale within months — always confirm current specs before making a purchasing decision.
• Enterprise pricing is not published, requiring a sales conversation to get an actual quote.

Security & Privacy

OpenAI states that ChatGPT Business and Enterprise plans do not train models on customers' business data by default. Enterprise adds SSO, SCIM provisioning, audit logs, a compliance API, and configurable data retention controls. For Free/Plus/Pro consumer plans, conversation data may be used to improve models unless a user disables model training in settings, or uses Temporary Chat, which is not used for training and isn't saved to chat history. Consult OpenAI's own privacy policy and enterprise trust documentation for full, current details before making a compliance decision — this article does not substitute for that.

Best For

People and teams who want one general-purpose AI assistant that also handles coding, research, documents, and images without switching tools — and who value the largest surrounding ecosystem of integrations, extensions, and community knowledge.

Alternatives

• Claude (Anthropic) — Anthropic's flagship assistant, often favored for writing and coding.
• Google Gemini — deeper integration with Google Workspace and Search.
• Microsoft Copilot — built into Microsoft 365 and Windows.
• Perplexity — search-first, citation-heavy alternative for research questions.
• Grok (xAI) — built into X (Twitter).
• DeepSeek, Mistral Le Chat — lower-cost/open- weight-friendly alternatives.

Frequently Asked Questions

Is ChatGPT free to use?

Yes — there's a free plan with access to a standard model, web search, and limited image generation, though it doesn't include the flagship reasoning model and may show ads in some markets.

What's the difference between ChatGPT Plus and Pro?

Plus ($20/month) unlocks the flagship reasoning model and core features like Projects, Deep Research, and Codex. The Pro tiers ($100 and $200/month) raise usage limits substantially and give heavier access to Deep Research and Codex for people who hit Plus's limits regularly.

Does ChatGPT have an API?

Yes. The OpenAI API is billed separately from ChatGPT's consumer/business plans, priced per input/output token and varying by model.

Can I use ChatGPT without an account?

OpenAI has offered limited logged-out access in some markets historically, but full functionality — including memory, history, and higher limits — requires a free account.

Final Summary

ChatGPT is OpenAI's flagship AI assistant, built on the GPT model family (GPT-6 Astra as of September 2026, with the GPT-5.6 family still powering lower tiers). It covers conversational chat, web search, vision, image generation, code execution, and agentic research/task automation in one product, priced from free through a $200/month Pro tier, with separate Business, Enterprise, and usage-based API pricing. Its main strengths are breadth of features and the largest surrounding ecosystem of any AI assistant; its main caveats are a fast-moving model/pricing lineup and no native Linux app.