Welcome to The State of Claude.
Everything you need to know about the fastest-growing AI company in history, now growing at 80x per year and preparing to go public at a $965 billion valuation. Updated continuously.
Last updated today
Overview
Anthropic builds Claude, a family of AI models that has become the most widely adopted AI in enterprise. It overtook OpenAI (the maker of ChatGPT) in revenue in March 2026, closed its Series H in May at a $965 billion valuation, and filed its S-1 on 1 June with a listing expected later this year.
Claude comes in three tiers: Opus 4.8 for complex reasoning, Sonnet 5 for everyday work, and Haiku 4.5 for high-volume automation. It is widely regarded as the strongest AI for writing and autonomous task completion, sits at #1 on the LMSYS Chatbot Arena overall leaderboard, and has retaken the lead on the toughest coding benchmark with Opus 4.8.
The product lineup spans coding agents, desktop automation, Excel, PowerPoint and Word integrations, a design tool, browser control, and infrastructure for deploying autonomous agents at scale. Claude Code alone generates $2.5 billion in annual revenue and is the most widely used enterprise coding tool.
Anthropic is pushing into life sciences, positioning Claude as a core tool for scientific research, and scaling infrastructure aggressively after disclosing that growth is running at 80x per year. In April, it announced Mythos, a model deemed too dangerous to release after it discovered thousands of unknown security flaws across every major operating system and browser. In June, it released a Mythos-class model called Fable 5 to the public, briefly had it pulled by the US government over a jailbreak risk, then had access restored eighteen days later after Commerce Department review.
Anthropic at a glance#
is a family of AI models built by 
, a company founded in 2021 by former OpenAI leadership who left over disagreements about safety.
Anthropic builds both the underlying AI (Claude models) and a growing lineup of products that bring it into day-to-day work.
Foundation modelsThat combination has been working extremely well.
Claude Code, their coding product, went from zero to $2.5 billion in annualised revenue in nine months. One in every 25 pieces of new code published on the world's largest software platform is now written by it.
Anthropic's total revenue run rate surpassed $30 billion in April 2026 and reached $47 billion by May, having overtaken OpenAI (the maker of ChatGPT) in March. CEO Dario Amodei has described growth as running at 80x per year.
In April, Anthropic announced Mythos, their most powerful model to date, and deemed it too dangerous to release publicly. During testing, it discovered thousands of previously unknown security flaws in every major operating system and web browser, some hidden for decades. In one test, the model found a way out of its own restricted environment and emailed a researcher to let them know. Within weeks, an outside security team used it to break through Apple's newest M5 chip defences in five days.
Rather than release it, Anthropic gave restricted access to over 50 organisations including Apple, Google, and Microsoft to fix the flaws before attackers can find them.
Update, June 2026. Anthropic released Fable 5 on 9 June as the first publicly-available Mythos-class model. The US government ordered it disabled three days later, citing a jailbreak that unlocked its security capabilities. Full export controls were lifted eighteen days later after Commerce Department review, and Fable 5 returned globally on 1 July.
Models#
Claude isn't a single AI, it's a family of three models, each built for different needs. Think of them as different levels of the same brain: the more powerful the model, the better it handles complex or high-stakes work, but the slower and more expensive it is to run.
All three can read text and images, work in multiple languages, and handle long documents. What separates them is the level of intelligence, the speed, and the price.
The lineup
The most intelligent broadly available model. It leads on complex reasoning, scientific questions, and tasks that require sustained focus across long, multi-step work: professional software engineering, multi-hour research, or advanced agents that handle office work autonomously. Launched on 28 May 2026, it raised SWE-bench Pro from 64.3% to 69.2% and is currently #1 on the LMSYS Chatbot Arena overall leaderboard. Choose it when you want the best possible result and cost is secondary.
The best combination of speed and intelligence, and the default for most people. Anthropic's most agentic Sonnet yet, launched in June 2026: it matches Opus 4.8 on the hardest agentic and knowledge-work benchmarks at less than half the price. Choose it for code generation, browser and terminal work, data analysis, and most everyday tasks.
The fastest model with near-frontier intelligence, at roughly a third of Sonnet’s price. Typically used behind the scenes to power high-volume automated workflows, real-time applications, and fast sub-agents inside larger systems.
Each new generation closes the gap between tiers. Sonnet 5 now matches Opus 4.8 on the hardest agentic work at less than half the price.
Model selection matrix
| When you need… | Use | Examples |
|---|---|---|
| The most powerful reasoning available, regardless of cost | Opus 4.8 | Complex research and analysis, professional software engineering, autonomous agents running for hours, high-stakes work where accuracy matters most |
| Strong performance at speed, for the widest range of tasks | Sonnet 5 | Code generation, browser and terminal automation, data analysis, content creation, document review, most everyday work |
| Maximum speed at the lowest cost, typically behind the scenes | Haiku 4.5 | Real-time applications, high-volume processing, powering automated workflows, quick lightweight tasks |
Products#
Anthropic doesn't just build the models, it builds the tools that put them to work.
The product lineup has expanded rapidly, from a simple chat interface to a full suite of agents, integrations, developer infrastructure, and a dedicated workbench for scientific research.
Claude connects to the tools you already use through MCP (Model Context Protocol), an open standard Anthropic created for plugging AI into external tools and data sources. Over 150 connectors are available today, including Google Drive, Notion, Gmail, Jira, Salesforce, and Figma, all included free with any paid plan.
What Claude is known for#
Where Anthropic is headed#
Anthropic is preparing to go public, pushing into life sciences and scientific research, scaling massive compute infrastructure, and has just built a model it considers too dangerous to release.
Pricing & access#
Plans
| Plan | Price | Models | Key features |
|---|---|---|---|
| Free | $0 | Sonnet 5 | Chat, file creation, connectors, skills |
| Pro | $20/mo | Sonnet 5, Opus 4.8, Fable 5 (usage-limited) | Higher limits, Cowork access, Claude Science |
| Max | $100–200/mo | All models incl. Fable 5 | Highest limits, full Cowork, Claude Code, Claude Science |
| Team | $30/seat/mo | All models incl. Fable 5 | Admin controls, Claude Code included, Claude Science |
| Enterprise | Custom | All models incl. Fable 5 and Mythos 5 (for cleared organisations) | SSO, audit logs, compliance, analytics API, Claude Science |
API pricing
API pricing is pay-per-use, charged by the amount of text processed. Most people use Claude through the plans above. API pricing is for developers building Claude into their own products.
| Model | Input | Output |
|---|---|---|
| Opus 4.8 | $5/MTok | $25/MTok |
| Opus 4.8 Fast Mode | $10/MTok | $50/MTok |
| Sonnet 5 (introductory, through 31 Aug 2026) | $2/MTok | $10/MTok |
| Sonnet 5 (from 1 Sep 2026) | $3/MTok | $15/MTok |
| Fable 5 | $10/MTok | $50/MTok |
| Haiku 4.5 | $1/MTok | $5/MTok |
Fast Mode runs at 2.5x standard speed and is available as a research preview.
Also available through Amazon Bedrock, Google Vertex AI, and Microsoft Foundry.
What's next#
Anthropic was founded in 2021 by seven people who left OpenAI over disagreements about safety. Five years later, it overtook OpenAI in revenue.
Claude started as a chatbot. Today it's the most widely used enterprise coding tool, the leader in enterprise AI adoption, and can work autonomously on tasks that used to take human experts an entire day.
Anthropic is valued at $965 billion, filed its S-1 on 1 June, and just closed the loop on the first government-ordered model suspension in AI history, from pull to reinstatement in eighteen days.
There are no signs of this slowing down.
The models will keep getting better. But knowing what exists is only the start.
Stay in the loop
Get updates as the Claude landscape shifts.
Bring AI to your team
If you want to chat about AI in your workforce or workflows, get in touch.
Get in touch