Find Claude’s official login, explore free access, compare pricing, and see how it stacks up against ChatGPT. Discover what’s new - Claude Fable 5.2 and Mythos 5.2, new Claude Enterprise security and admin features, Claude Duet and the Claude 5.5 models.
NewClaude Fable 5.2 & Mythos 5.2 are out, and Claude Enterprise has new security and admin features →Claude is a general-purpose AI assistant: you give it text, images or files and it writes, reasons, codes, analyses and - increasingly - acts on your behalf. It is made by Anthropic, an AI safety company founded in 2021, and trained with a method the company calls constitutional AI.
Over the past two years Claude's centre of gravity has shifted from "chatbot that writes well" to "model that can run a long task for you". That is why the current line-up spans four tiers - Haiku for speed, Sonnet for everyday work, Opus as the flagship and Fable (now Fable 5.2) for the hardest technical problems - and why it ships as a terminal agent, an API and a proactive assistant as well as a chat app.
Every current Claude model shares a 1M-token context window, 128k output and a June 2026 knowledge cutoff, so the choice comes down to capability against cost and speed. Anthropic's documentation recommends starting with Opus 5.5; Sonnet 5.5 does most of the same work at half the price; Haiku 5.5 is 40× cheaper again; and Fable 5.2 - released 9 November 2026 at a 20% lower price than Fable 5.1 - is for the hardest coding, maths, science and engineering work.
claude-fable-5-2claude-opus-5-5claude-sonnet-5-5claude-haiku-5-5Still callable: Fable 5, Opus 5, Sonnet 5, Haiku 4.5, Opus 4.8 / 4.7 / 4.6 and Sonnet 4.6. Sonnet 4.5 retires from the API on 30 November 2026. Mythos 5.2 - the same model as Fable 5.2 with fewer safeguards - is limited to Anthropic's Cyber and Life Sciences Verification Programs and Claude Security on Enterprise plans.
Claude is no longer just a chat window. The same models power a terminal agent, an API, a browser extension, a design canvas and a proactive assistant. Here is the full line-up.
See all products →Claude Code reads your codebase, plans a change, edits files, runs the tests and opens the pull request - from the shell, your editor or the browser. Give it a task, step away, and review the diff when it is done.
One endpoint, four models, billed per million tokens. Call it from the Python or TypeScript SDK or plain HTTPS, or through AWS Bedrock, Google Cloud Vertex AI and Microsoft Foundry. Prompt caching and the Batch API are the two levers that actually change the bill.
import anthropic
client = anthropic.Anthropic()
msg = client.messages.create(
model="claude-opus-5-5",
max_tokens=1024,
messages=[{"role": "user",
"content": "Summarise this contract…"}],
)
print(msg.content[0].text)| Model | Input | Output | Cache read | Batch in / out |
|---|---|---|---|---|
| Fable 5.2 | $8 | $40 | $0.20 | $4 / $20 |
| Opus 5.5 | $4 | $20 | $0.20 | $2 / $10 |
| Sonnet 5.5 | $2 | $10 | $0.10 | $1 / $5 |
| Haiku 5.5 ≤100k | $0.10 | $0.50 | $0.01 | $0.05 / $0.25 |
Per million tokens, standard tier. Haiku 5.5 prompts over 100k tokens are billed at $0.50 / $2.50. Fable 5.2 cut Fable rates by 20% (from $10 / $50); its fast mode runs up to 2.5× faster at $16 / $80, and US-only inference costs 1.1×. Opus 5.5 cut every Opus rate by 20%.
Subscriptions pay for the apps; the API is billed separately per token. An API key does not give you Pro, and Pro does not give you API credit - though Max, Ultra and Team now include a monthly API allowance.
Trying Claude; light daily use
$0
Individuals using Claude most days
$20per month · $200/year on annual
Heavy daily users hitting Pro limits
$100per month · monthly only
All-day agentic coding
$200per month · monthly only
2–150 people with admin controls
$25per seat/month · Premium seat $125 ($100 annual)
Governance, compliance, scale
$20per seat/month + usage at API rates · annual, 20-seat min.
Every plan is metered by a five-hour window and a weekly cap rather than a message count - see how usage limits work. Prices exclude tax; confirm current figures at claude.com/pricing.
Official Claude apps exist for Mac, Windows, Linux, iPhone, iPad, Android and ChromeOS, plus a Chrome extension and the Claude Code terminal agent. Every link here points to Anthropic's own stores and download pages.
claude in any project folder.Claude works best when you treat it like a capable colleague: give it the goal, the material and the format you want back, then let it run. Here is what experienced users do differently.
Most “Claude is broken” moments are one of these. Work through the diagnostic order on the right - status, connection, limits, browser, app, account - and nine times out of ten you will find it before you need support. Region and sign-in problems have their own sections: availability and login.
claude and choose “Log in” again, make sure you picked the account that holds the paid plan, and update to the latest version - old builds can lose their token after a plan change.Claude Code →Still stuck? The full troubleshooting guide covers error codes, file-upload failures and API-specific issues, and Anthropic's own help centre is at support.claude.com.
Anthropic does not publish a message count for any plan. What it publishes is a mechanism - a five-hour session window, a weekly cap, and relative multipliers between plans - and once you understand it, the “limit reached” message stops being a surprise.
Claude.ai and the API share one supported-regions list, now covering 175+ countries and territories, but individual products roll out at different speeds and a few are limited to the United States.
| Product | Available in | Notes |
|---|---|---|
| Claude.ai (Free, Pro, Max, Ultra) | 175+ countries and territories | Same list as the API. Sign-up is blocked, not just paid plans, where a country is unsupported. |
| Claude API & Console | 175+ countries and territories | Also reachable through AWS Bedrock, Google Cloud Vertex AI and Microsoft Foundry, each with its own region list. |
| Claude Code, Cowork, Chrome extension | Wherever a paid plan is sold | Need Pro or higher; the Chrome extension installs from the Chrome Web Store. |
| Team & Enterprise | Most supported countries | Billing in USD; EUR pricing shown in parts of Europe. Enterprise data-residency options vary by cloud. |
| Claude Duet (research preview) | United States only | Waitlist for Pro, Max and Ultra subscribers; Team and Enterprise not yet included. |
| Fable 5.2 | All paid plans and platforms | Pro, Max, Ultra, Team and Enterprise, the Claude Platform, AWS, Google Cloud and Microsoft Foundry. Fast mode on the Claude Platform, Google Cloud, Microsoft Foundry and Claude Code. |
| Mythos 5.2 | Verification programmes | Only for organisations approved under the Cyber or Life Sciences Verification Program, or through Claude Security on Enterprise plans. Fable 5.2 is the public equivalent. |
Two questions come up before anyone uploads a contract or a customer list: does Anthropic train on it, and how long is it kept? The answer depends on which plan you are on - and then there are the practical limits on what you can upload at all.
Attach files to a single chat, or add them to a Project's knowledge base so every new conversation starts with them loaded. Chat uploads are the roomier of the two; Project files are smaller but unlimited in number, as long as the extracted text fits the context window.
*XLSX needs file creation and analysis enabled in Settings. Scanned PDFs without a text layer are read as images, which costs more context. Files count toward usage: a 300-page PDF spends far more of your five-hour window than a short question.
Claude lives on a handful of Anthropic-owned domains. Knowing which is which keeps you off phishing pages and saves a support ticket when you cannot find your billing or your API key.
Anthropic's Opus 5.5 and OpenAI's GPT-6 Astra are the two flagships most people choose between. On published numbers Claude leads on measured intelligence and API price while ChatGPT wins on generative media and plan variety - and on most everyday tasks the honest answer is that either will do the job.
Two big announcements this month - Claude Fable 5.2 and Mythos 5.2, and new security and admin features for Claude Enterprise - plus two more changes worth knowing about, the full details and the release timeline. The updates page tracks every model, price and feature change as it happens.
Anthropic's most capable models for technical work, with gains in coding, reasoning and vision. Output is up to 30% faster than the 5.1 versions and Fable 5.2 is 20% cheaper at $8 / $40 per million tokens. Fable 5.2 is available today on every paid plan, the Claude Platform and all major clouds; Mythos 5.2 - the same model with fewer safeguards - goes to vetted cyber and life-sciences organisations.
Full Fable 5.2 details ↓A security center that checks your settings against Anthropic's recommended configuration, generally available inference hooks for your existing data-security tools, centralized management for company-wide settings, and new access controls: Admin API sign-out, a self-serve IP allowlist, Claude Desktop management with shadow-IT controls, and EDR sensors inside Cowork. These build on the Compliance API's 28 security-tool integrations, custom roles, SCIM, audit logs and the HIPAA-ready option.
Full feature list ↓A shared Claude teammate inside Slack. Tag it in a channel or project thread and it works under its own identity, remembers channel context, schedules its own follow-ups and can pursue a task over hours or days. Admins scope it per channel, set spend limits and review logs of every action. Replaces the old Claude in Slack app; wider platform support to follow.
Claude Tag →Spotted in the Claude iOS app under the working name "Claude Money" and renamed Personal Finance in the latest build: connect bank accounts through Plaid and let Claude analyse spending and build dashboards on your own data. Anthropic has not announced it, so scope and availability are unconfirmed - treat it as in testing. Third-party finance connectors such as PocketSmith already work today.
Personal Finance →Fable and Mythos are the same underlying model. Fable 5.2 carries cybersecurity and biology safeguards so it can be sold to everyone; Mythos 5.2 runs with fewer of them for approved organisations. Anthropic positions Fable 5.2 for hard problems in engineering, mathematics, science and robotics, and keeps Opus 5.5 as the better fit for everyday tasks.
claude-fable-5-2 · 1M context · 128k output · US-only inference at 1.1×| Benchmark (Anthropic) | Fable 5.2 | Fable 5.1 | Opus 5.5 |
|---|---|---|---|
| FrontierCode 2.0 (agentic coding) | 49.1% | 32.4% | 45.5% |
| FrontierSWE v2 (agentic coding) | 69.9% | 56.3% | 62.3% |
| AutomationBench (business workflows) | 51.4% | 31.4% | 42.5% |
| Terminal-Bench-Science 0.1 | 74.6% | 52.6% | 58.7% |
| OSWorld 2.1 (computer use) | 89.3% | 83.5% | 87.2% |
| RoboDojo Sim (robotics) | 64.9% | 27.6% | 39.4% |
| Remote Labor Index | 41.3% | 17.9% | 21.3% |
| Humanity's Last Exam (with tools) | 74.2% | 65.6% | 67.7% |
Source: Anthropic - Claude Fable 5.2. Benchmarks are Anthropic's own; Fable 5.2 also leads GPT-6 Astra on most rows (69.9% vs 65.5% on FrontierSWE v2).
Seven additions give security teams one view of how Claude is configured, what it does on company devices and who can reach it. Inference hooks are generally available; everything else is in beta for Claude Enterprise, with some beta exceptions listed in Anthropic's Help Center.
Builds on the existing Compliance API (28 security and compliance tool integrations), custom roles with per-group feature and spend limits, SCIM, audit logs, the HIPAA-ready Enterprise option and the Enterprise Analytics API. Source: Claude blog - More visibility and control over Claude Enterprise for security teams.
Short answers to the questions people search for most, each pointing to the section with the detail.
Claude Fable 5.2 and Claude Mythos 5.2, released 9 November 2026, are the newest. The current lineup is Fable 5.2, Opus 5.5, Sonnet 5.5 and Haiku 5.5. Fable 5.2 is for the hardest technical work; Opus 5.5 remains the recommended everyday default. Compare the models →
Yes. The free plan runs on Sonnet 5.5 and Haiku 5.5 with web search, memory, file creation, Artifacts, Design, Slides, Docs and up to five Projects. Opus, Fable, Claude Code, Cowork, Research and the Chrome extension need a paid plan. See every plan →
$20 a month, or $200 a year ($17 a month). Max is $100, Ultra is $200, Team seats are $25 or $125 a month, and Enterprise is $20 per seat plus usage. Plan details →
Anthropic does not publish a number. Usage is a rolling five-hour window plus a weekly cap, metered in tokens, with Fable counted separately. Max is 5× Pro; Ultra is 10× Pro per week with no five-hour limit when verified. How limits work →
They are capability tiers at different prices: Haiku 5.5 is fastest and cheapest, Sonnet 5.5 is the everyday workhorse, Opus 5.5 is the flagship, and Fable 5.2 is the top-end model for the hardest technical work. All share a 1M context. Model comparison →
No. Claude has no native image, video or music generation and cannot take audio or video input. It draws charts, diagrams and designs as code instead. What Claude can do →
Yes - iPhone, iPad, Android, ChromeOS, macOS 11+, Windows 10+, a Linux beta, a Chrome extension, Microsoft 365 add-ins and the Claude Code terminal agent. Download links →
On consumer plans only if you opt in (Settings → Privacy); Team, Enterprise and API customers are never trained on under the Commercial Terms. Privacy details →
Every paid plan - Pro, Max, Ultra, Team and Enterprise - with usage drawn from the same pool as chat. About Claude Code →
A research preview of a proactive-agent app that briefs you each morning and continues tasks in the cloud. It runs on Opus 5.5, is free with Pro, Max and Ultra during the preview, and is US-only by waitlist. All products →
PDF, DOCX, TXT, HTML, ODT, RTF, EPUB, JSON, CSV, XLSX and PNG/JPEG/GIF/WebP images - up to 20 files of 500 MB per chat and 1,000 pages per PDF. No video, audio, executables or archives. Upload limits →
Anthropic's most capable generally available model, released 9 November 2026. It costs $8 / $40 per million tokens (down from $10 / $50), generates output up to 30% faster than Fable 5.1, and is available on Pro, Max, Ultra, Team and Enterprise plans, the Claude Platform and the major clouds. Its twin, Mythos 5.2, is the same model with fewer safeguards for vetted cyber and life-sciences organisations. Fable 5.2 details →
Announced 5 November 2026: a security center that checks your settings against Anthropic's recommended configuration, generally available inference hooks for existing data-security tools, centralized management at admin.claude.com, member sign-out through the Admin API, a self-serve IP allowlist, Claude Desktop management with shadow-IT controls, and EDR sensors inside Cowork. Everything except inference hooks is in beta. Full feature list →
No. This is an independent reference. The official product is at claude.ai, the company at anthropic.com and developer docs at platform.claude.com; we never ask you to sign in here. Official sites →