Claude is Anthropic's family of AI models and the assistant apps built on them. You can chat with it at claude.ai and in the desktop and mobile apps, use it as a coding agent through Claude Code, or build it into your own software through the Claude API. Anthropic is an AI safety and research company based in San Francisco.
As of October 2026 there are four current models: Claude Fable 5.1, the most capable generally available model; Claude Opus 5.5, the recommended starting point for most API work; Claude Sonnet 5.5, the best balance of speed and intelligence; and Claude Haiku 5.5, the fastest and cheapest. All four have a 1-million-token context window.
There's a free plan, and paid plans start with Pro at $20 per month, which adds Opus and Claude Code. Against ChatGPT, Claude is especially strong at coding, long documents and careful writing, while ChatGPT has more built-in consumer features such as image and video generation.
Three ways to use Claude
The models read text and images and write text. They can reason through problems, write and debug code, analyse long documents, use tools, and act as agents that complete multi-step tasks. How you reach them depends on what you're doing:
- The Claude apps: the web at claude.ai, desktop apps for macOS and Windows, and iOS and Android apps.
- Claude Code, an agentic coding tool that works in your terminal, IDE, desktop app or browser.
- The Claude API, for building Claude into your own apps. It's also available through Amazon Bedrock, Google Cloud and Microsoft Foundry.
Which Claude model should you pick?
Anthropic's models overview gives refreshingly simple advice: start with Opus 5.5, and move to Fable 5.1 only when Opus at higher effort still falls short. I'd follow it. Most of the decision is really about cost and latency.
Model | Best for | API ID | API price per 1M tokens (input / output) | Speed |
|---|
Claude Fable 5.1 | Demanding reasoning and long-horizon agentic work | claude-fable-5-1
| $10 / $50 | Slower |
Claude Opus 5.5 | Long-running agentic coding and knowledge work | claude-opus-5-5
| $4 / $20 | Moderate |
Claude Sonnet 5.5 | Everyday work with the best speed-to-intelligence balance | claude-sonnet-5-5
| $2 / $10 | Fast |
Claude Haiku 5.5 | High-volume tasks such as classification, extraction and routing | claude-haiku-5-5
| From $0.10 / $0.50 | Fastest |
All four share the same core specs: a 1M-token context window, up to 128K output tokens, text and image input with text output, and a reliable knowledge cutoff of June 2026. They also share adaptive thinking, where the model decides how much to reason and you steer it with an effort setting from low to max.
Fable 5.1 arrived on September 1, 2026, as the follow-up to Claude Fable 5. Anthropic built it for ambitious, long-running projects, and it thinks more and takes longer per request, so it suits deep work rather than quick chat. Anthropic also runs a more restricted line, Claude Mythos, which is available only through a limited access program.
Opus 5.5 replaced Claude Opus 5 at a lower API price and is Anthropic's recommendation for most workloads. Thinking is always on, but its default effort is medium. Leaving it there for complex coding is a common mistake; raise it to high or xhigh.
Sonnet 5.5 is the everyday model, and a strong default for chat apps, agents that make many tool calls, and coding assistants where latency matters.
Haiku 5.5 is the model for scale. Its "from" price applies to prompts up to 100K tokens; longer prompts cost more. It's widely distributed: GitHub added Claude Haiku 5.5 to Copilot, and it's also available on Amazon Bedrock.
Older models such as Claude Fable 5, Opus 5, Opus 4.8 and Sonnet 5 remain available as legacy options. Old IDs keep working until retirement, but hardcoding one means you miss newer and often cheaper models, so new projects should use the current four.
Claude plans, and the Fable exception
These are the plans on Claude's pricing page as of October 2026. Anthropic notes that prices and limits can change.
Plan | Price | What you get |
|---|
Free | $0 | Sonnet and Haiku, up to 5 projects, connectors. Limits reset on a rolling 5-hour window. |
Pro | $20/month, or $17/month billed annually | Adds Opus, Claude Code, Research, unlimited projects, and at least 5x Free usage |
Max | From $100/month | 5x or 20x Pro usage, higher output limits, priority access at busy times, Fable included in weekly limits |
Team | Standard seat $25/month ($20 annual); Premium seat $125/month ($100 annual) | 2 to 150 users, central billing, SSO, connector admin controls, usage analytics |
Enterprise | $20/seat/month billed annually (20-seat minimum) plus usage at API rates | Adds role-based access, SCIM, audit logs, Compliance API, custom retention, IP allowlisting, HIPAA-ready option |
Fable doesn't follow the normal plan limits, and this catches Pro subscribers out. According to Anthropic's support article on Fable, Max and premium Team or Enterprise seats can spend up to 50% of their weekly usage limit on Fable at no extra cost, though Fable burns through that limit faster than other models. On Pro and standard seats, Fable isn't covered by plan limits at all; it runs on pay-as-you-go usage credits, if you've enabled them. Free has no Fable access.
What the Claude app can do
The core app is a chat interface with file uploads, projects (shared files and instructions for a set of chats), web search, and, on paid plans, Research mode for longer cited reports.
Anthropic launched Claude Cowork, an agent mode that works with your files and apps, in early 2026. Reports from September 2026 say it's being folded into the regular app, and the pricing page now says "Claude Cowork is now just Claude." In practice, you ask for a deliverable in a normal chat and Claude works through the steps.
Connectors link Claude to tools such as Google Drive and Slack. They're built on the Model Context Protocol (MCP), the open standard Anthropic created for connecting AI to data sources. Pro and Max also include Claude in Chrome and Claude in Microsoft 365.
Getting started with Claude Code
Claude Code is Anthropic's agentic coding tool. It reads your codebase, edits files, runs commands and works with git. It runs in the terminal, in VS Code and Cursor, in JetBrains IDEs, in the Claude desktop app, and on the web at claude.ai/code. You need a paid Claude subscription or an Anthropic Console account.
On Windows PowerShell:
irm https://claude.ai/install.ps1 | iex
On macOS, Linux or WSL:
curl -fsSL https://claude.ai/install.sh | bash
You can also use winget install Anthropic.ClaudeCode or brew install --cask claude-code. On Windows, CMD works too, and installing Git for Windows is recommended so Claude Code can use Bash; without it, Claude Code falls back to PowerShell. Then start it inside a project:
cd your-project
claude
Developers use it to write tests, run them and fix the failures; to trace a bug from an error message to its root cause; and to make changes across many files, then commit and open a pull request. It also runs in CI through GitHub Actions or GitLab CI/CD for code review and issue triage, and it connects to Jira, Google Drive or your own tools through MCP servers.
The first thing I'd set up is a CLAUDE.md file at the repository root with your project's coding standards. Claude Code reads it at the start of every session, and it can also read an existing AGENTS.md. The Claude Code documentation covers the rest: skills, hooks, subagents and scheduled tasks.
Calling the Claude API
The API centres on one endpoint, the Messages API: you send a list of messages and get a response. Create an account at platform.claude.com, generate an API key, store it in the ANTHROPIC_API_KEY environment variable, install the official SDK for your language, and send a request with a model ID from the table above.
For .NET, Anthropic publishes an official C# SDK on NuGet as the Anthropic package:
dotnet new console -n ClaudeDemo
cd ClaudeDemo
dotnet add package Anthropic
using Anthropic;
using Anthropic.Models.Messages;
AnthropicClient client = new(); // reads ANTHROPIC_API_KEY
var response = await client.Messages.Create(new MessageCreateParams
{
Model = "claude-opus-5-5",
MaxTokens = 16000,
Messages = [new() { Role = Role.User, Content = "Explain records vs classes in C# with one example each." }]
});
foreach (var text in response.Content.Select(b => b.Value).OfType<TextBlock>())
{
Console.WriteLine(text.Text);
}
The Python version looks much the same. This one uses Sonnet at low effort, which is a sensible setting for a summarisation job:
import anthropic
client = anthropic.Anthropic() # reads ANTHROPIC_API_KEY
message = client.messages.create(
model="claude-sonnet-5-5",
max_tokens=16000,
output_config={"effort": "low"},
messages=[{"role": "user", "content": "Summarise this changelog in 5 bullets: ..."}],
)
for block in message.content:
if block.type == "text":
print(block.text)
API habits that save time and money
If you've used older Claude models, two things have changed. The current models use adaptive thinking, so you control cost with output_config.effort; the old budget_tokens setting now returns an error. Sampling parameters are rejected too, so don't send temperature or top_p.
Beyond that, prompt caching makes repeated system prompts and documents much cheaper on later requests, and the Message Batches API is 50% cheaper for anything that isn't urgent. Check stop_reason before you read the content; a value of refusal means a safety classifier declined the request.
If you run on a cloud provider, note that Amazon Bedrock prefixes the ID, for example anthropic.claude-opus-5-5, while Google Cloud and Microsoft Foundry use the same IDs as the Claude API.
Claude or ChatGPT?
Both are excellent general assistants, and the differences show up in specific jobs.
Area | Claude | ChatGPT |
|---|
Top model (Oct 2026) | Claude Fable 5.1 | GPT-6 Astra |
Entry paid plan | Pro, $20/month | Plus, $20/month (Go is about $8) |
Coding agent | Claude Code | Codex |
Agent for office work | Built into Claude (formerly Cowork) | ChatGPT Work |
Image and video generation | Not a core feature | Built in |
Context window (API) | 1M tokens on all current models | About 1.05M tokens on GPT-6 models |
Business plans | Team, Enterprise | Business, Enterprise |
Claude is the better fit if you write a lot of code and want a strong terminal and IDE agent, if you work with long documents, contracts or codebases that need careful reading, or if you want writing that needs less editing. ChatGPT makes more sense if you want images, voice and video generation in one app, if your team has standardised on OpenAI tools, or if you need a cheap entry plan like Go.
On coding specifically, Claude Code is one of the most popular coding agents, but ChatGPT's Codex and GPT-6 models are strong too, and results vary by language and task. Plenty of developers use both. The honest test is to run your own real tasks through each for a week.
Whichever you choose, keep API keys and passwords out of prompts and shared code samples, and treat the output as a draft: run the tests and check facts against sources before you rely on it.
Join the conversation! Your thoughts help the community grow.