Anthropic has released more than a dozen Claude models in 2026 so far. Over the same months, Claude grew from a chat app into a tool that carries out multi-step work. This guide covers every major Claude release from January to late October, and what each one changed for people using Claude at work.
The Claude models available today
These are the current models in the Claude API, with standard prices per million tokens in US dollars. All four have a 1 million token context window.
| Model | Best for, in Anthropic’s words | Input | Output |
|---|---|---|---|
| Fable 5.1 | Demanding reasoning and long agentic work | US$10 | US$50 |
| Opus 5.5 | Long-running agentic coding and knowledge work | US$4 | US$20 |
| Sonnet 5.5 | The best mix of speed and intelligence | US$2 | US$10 |
| Haiku 5.5 | High-volume tasks such as sorting and extraction | US$0.10 | US$0.50 |
Haiku 5.5 costs more for prompts over 100,000 tokens: US$0.50 for input and US$2.50 for output. The Batch API takes 50% off input and output prices for work that can wait.
Claude API output price per million tokens (US$)
January to March brought Cowork and longer context
Cowork launched as a research preview on 12 January for Max subscribers on macOS. It lets Claude work through tasks on your files inside an isolated virtual machine, rather than only answering in chat. Pro subscribers got access on 16 January.
Claude Opus 4.6 followed on 5 February. It was the first Opus model with a 1 million token context window, in beta. It also added agent teams in Claude Code, and Claude in PowerPoint launched as a research preview.
Claude Sonnet 4.6 arrived on 17 February. It became the default model for Free and Pro users in Claude and in Cowork.
Other changes in this period:
- Claude Code Security launched on 20 February as a limited research preview for Team and Enterprise customers. It scans code for security flaws and suggests fixes for a person to approve.
- Cowork plugins arrived on 24 February with a plugin marketplace for Team and Enterprise plans. Scheduled tasks followed on 25 February.
- Memory of past chats became available to all users, including Free, on 2 March.
- The 1 million token context window came out of beta for Opus 4.6 and Sonnet 4.6 on 13 March, at standard pricing.
- Dispatch arrived on 17 March. It keeps one conversation with Claude going across your phone and computer. On 23 March, Claude gained computer use in Cowork and Claude Code, as a research preview.
April to June brought Opus 4.7, Claude Design and Fable 5
Project Glasswing launched on 7 April with Claude Mythos Preview, a model shared only with selected partners for defensive security work.
Cowork became generally available on macOS and Windows on 9 April.
Claude Opus 4.7 arrived on 16 April with higher-resolution image input and a new “xhigh” effort level. A day later, Anthropic Labs launched Claude Design, a research preview for creating designs, prototypes and slides with Claude.
Claude Opus 4.8 followed on 28 May. It added an effort control in Claude and Cowork, available on all plans.
Claude Fable 5 launched on 9 June as Anthropic’s most capable model open to all customers. Its sibling, Mythos 5, went to a small group of Project Glasswing partners.
On 12 June, the US government applied export controls to both models, and Anthropic suspended access for all users. The controls were lifted on 30 June. Fable 5 returned globally on 1 July with new safeguards, and Mythos 5 returned for a set of US organisations.
Claude Sonnet 5 arrived on 30 June at US$2 and US$10 per million tokens, below Sonnet 4.6’s US$3 and US$15. On 10 August, Anthropic made that introductory price the standard price.
July to September brought Opus 5 and one Claude app
Cowork came to the web and mobile in beta on 7 July. Scheduled tasks now run with no device online.
Claude Opus 5 launched on 24 July and became the default model on Max plans.
In August, the computer use tool came out of beta in the Claude API (19 August). Memory began to work across chat and Cowork (25 August).
Claude Fable 5.1 arrived on 1 September at the same token prices as Fable 5. Anthropic estimates typical workloads cost about 25% less, mainly because reading cached input now costs 75% less.
On 16 September, Cowork and chat merged into one Claude. Any conversation can now take on multi-step work, and Claude Docs and Claude Slides launched in beta. The change reached Pro, Max and Ultra first. We cover it in Claude Cowork is now in every chat.
Claude Opus 5.5 followed on 22 September at US$4 and US$20 per million tokens, 20% less than Opus 5. Anthropic says it costs 40% less than Opus 5 on typical work. Claude Sonnet 5.5 arrived on 28 September at the same price as Sonnet 5.
October brought Haiku 5.5, Duet and a Singapore office
- Claude Haiku 5.5 (7 October) is the first Haiku model with an adjustable effort setting. On the same day, Anthropic announced monthly API credits for some paid plans: US$100 on Max, US$200 on Ultra and up to US$500 pooled on Team.
- Anthropic opened a Singapore office on 14 October, its fifth in Asia-Pacific. It also signed a memorandum of understanding with Singapore’s Ministry of Digital Development and Information.
- Claude Duet (21 October) is a research preview of an app where Claude works as a proactive agent. It runs on iOS, Android and claude.ai, with a waitlist that starts with US Pro, Max and Ultra subscribers.
- Claude Models on Google Cloud (Anthropic managed) became generally available on 22 October. Google Cloud customers get the full Claude Platform with Google Cloud billing.
- The combined chat and Cowork experience became generally available for Claude Enterprise on 23 October.
- Anthropic filed an S-1 with the US Securities and Exchange Commission on 26 October for a proposed share listing on Nasdaq.
What the Claude plans cost today
Claude’s individual paid plans are Pro, Max and Ultra. Pro costs US$20 a month, or US$17 a month billed annually. Max costs US$100 a month and Ultra US$200, and both include API credits each month. Team seats start at US$20 per user per month billed annually, or US$25 billed monthly.
A chat plan covers Claude in the app, including Cowork. Automations that run inside your own systems, such as a form that triggers an AI step, use the Claude API. The API is billed separately by usage, so the monthly credits on Max, Ultra and Team only go part of the way.
What to do next
- Move off Claude Sonnet 4.5 before 30 November. It retires from the Claude API on that date. Sonnet 5.5 is the recommended replacement, and it also costs less per token.
- Try the 1 million token context window on long documents, contracts or full reports.
- Look at scheduled tasks for weekly work such as reports and research digests.
References
28 official sources
- Pricing (Claude Platform docs)
- Models overview
- Claude Platform release notes
- Release notes
- Model deprecations
- Claude plans and pricing
- Introducing Claude Opus 4.6
- Introducing Claude Sonnet 4.6
- Making frontier cybersecurity capabilities available to defenders
- Put Claude to work on your computer
- Project Glasswing
- Introducing Claude Opus 4.7
- Introducing Claude Design by Anthropic Labs
- Introducing Claude Opus 4.8
- Claude Fable 5 and Claude Mythos 5
- Access to Claude Fable 5 and Mythos 5 is now restored
- Introducing Claude Sonnet 5
- Claude Cowork is coming to mobile and web
- Claude Opus 5
- Claude Fable 5.1 and Mythos 5.1
- Claude Cowork and chat are now one Claude
- Introducing Claude Opus 5.5
- Introducing Claude Sonnet 5.5
- Claude Haiku 5.5
- Anthropic opens Singapore office and signs MOU with the Government of Singapore
- Introducing Claude Duet
- The full Claude Platform, available on Google Cloud
- Anthropic files S-1 with the SEC for proposed initial public offering