Claude release timeline
Claude releases, dated and tracked
30 generally available models in 44 months, from Claude on Mar 14, 2023 to Claude Fable 5.2 on Nov 9, 2026. Plus Claude app, API, SDK, Claude Code, MCP and Skills updates. Each entry links to an official announcement or changelog.
Last reviewed · 78 entries · 44 product updates and milestones
Follow the official release feeds
Reviewed through Nov 12, 2026. Latest verified model release: Nov 9, 2026. Latest Claude Code release: Nov 10, 2026. This curated timeline highlights releases and material changes. The feeds include every patch and fix.
By the numbers
as of Nov 12, 2026models_released
30
plus 4 restricted Mythos models
avg_gap
53 days
between 26 release days
fastest_gap
6 days
Opus 5.5 to Sonnet 5.5
longest_wait
125 days
3.5 Sonnet (upgraded) + 3.5 Haiku to 3.7 Sonnet
opus_versions
9
6 in the last 12 months
sonnet_versions
9
the most of any tier
haiku_versions
6
counting Claude Instant
busiest_year
2026
12 models so far
models per year
- 20235
- 20246
- 20257
- 202612
Current lineup
per Anthropic's models overviewTimeline
newest first2026
12 models·Claude Code
Adds Vertex fast mode in global, US and EU regions, with standard-speed fallback.
·Claude Code
Makes Fable 5.2 the default Fable model and adds settings-file support for Remote Control.
·API & SDKs
Claude API fast mode offers up to 2.5× output speed at $16/$80 per million input/output tokens, without a beta header.
·API & SDKs
Always-on adaptive thinking, stricter conversation-prefix checks and updated computer-use requirements. Check compatibility before switching an existing integration.
·Mythos
Claude Mythos 5.2
Restricted accessThe corresponding model for organizations approved through Anthropic verification programs. Access remains restricted.
claude-mythos-5-21M context·Fable·33 days after the previous release
Claude Fable 5.2
CurrentA new frontier model for demanding reasoning, coding and extended agent work. Standard API pricing is $8 per million input tokens and $40 per million output tokens.
claude-fable-5-21M context·Claude Code
Adds background MCP tasks, latency optimization guidance and additional permission controls.
·API & SDKs
Adds security-posture data, IP allowlists and organization-member sign-out.
·Product
US users on Pro, Max and Ultra can connect financial accounts and ask about their finances. Available in beta.
·Product
A security center, inference hooks, centralized management and additional access controls arrive, with several features in beta.
·Claude Code
Adds mod activation consent records and better visibility into authentication expiration.
·Claude Code
Adds blocking failure behavior for Agent SDK hooks, alongside permission and organization-policy improvements.
·API & SDKs
Adds Mythos 5, Opus 4.7 and eligible Sonnet 4.6 output.
·API & SDKs
Server-held conversation state lets subsequent API requests send only new messages, reducing upload size and time to first token. Threads expire after five idle minutes.
·API & SDKs
Removal is scheduled for November 30. Migrate fast-mode requests to Opus 5.5.
·API & SDKs
New usage and cost endpoints add flexible filtering and grouping. Older reporting APIs remain supported.
·API & SDKs
First-party Messages and Token Counting requests now accept 128 MB.
·API & SDKs
Enterprise admins can manage Slack settings and access bundles.
·API & SDKs
Google IAM authentication and Marketplace billing; Opus 5.5 fast mode.
·Milestone
Anthropic donates the open Agent Skills standard to the AAIF under the Linux Foundation.
·Product
A proactive Claude app enters research preview for Pro and Max subscribers.
·API & SDKs
API version 2026-10-21 ships with new major SDK and CLI versions. Review the migration guide for breaking changes.
·API & SDKs
Hooks graduate; user profiles, beta webhooks and web-fetch controls arrive.
·API & SDKs
Adds between-tools thinking discovery and saved user instructions.
·API & SDKs
Fast mode leaves research preview for Opus 5.5 and becomes available on Azure-hosted Microsoft Foundry deployments.
·API & SDKs
Scheduled deployments reach Claude Platform on AWS.
·API & SDKs
Trace API requests and Managed Agents sessions in external observability tools. SDK tracing covers seven languages.
·API & SDKs
Agents can orchestrate phased workflows across multiple agents, with background execution and progress events. Available in beta.
·API & SDKs
Max, Ultra and Team plans gain monthly API credits. Sonnet 5.5 cache-read pricing is halved.
·Haiku·9 days after the previous release
Claude Haiku 5.5
CurrentA new small model for fast, high-volume work, with adjustable effort and support for coding subagents.
claude-haiku-5-51M context·API & SDKs
Discover supported server tools through model capabilities.
·API & SDKs
Model capabilities now report whether thinking can be disabled.
·Sonnet·6 days after the previous release
Claude Sonnet 5.5
CurrentThe second Claude 5.5 model. Over 30% faster than Sonnet 5 and up to 30% cheaper per task. The fastest Sonnet yet.
claude-sonnet-5-51M context·Product
One catalog for 2,000+ connectors and plugins, Claude-powered products, and service partners.
·Opus·21 days after the previous release
Claude Opus 5.5
CurrentThe first Claude 5.5 model. Performs at the level of Fable 5.1 on most work and costs 40% less to run than Opus 5.
claude-opus-5-51M context·Mythos
Claude Mythos 5.1
Restricted accessThe same model as Fable 5.1 with different safeguards, limited to trusted access programs.
claude-mythos-5-1·Fable·39 days after the previous release
Cheaper to run than Fable 5 and better at coding and knowledge work, with research skills Anthropic calls an early look at AI-driven science.
claude-fable-5-11M context·Opus·24 days after the previous release
Comes close to Fable 5 at half the price.
claude-opus-5·Milestone
With the export controls lifted, Fable 5 is available again globally, with new classifiers that block more cybersecurity tasks.
·Sonnet·21 days after the previous release
"The most agentic Sonnet model yet", with near-Opus performance at Sonnet prices.
claude-sonnet-5·Milestone
Three days after launch, US export controls on Fable 5 and Mythos 5 force Anthropic to restrict access.
·Mythos
Claude Mythos 5
Restricted accessThe same generation as Fable 5 with fewer safeguards, offered only through trusted access programs.
claude-mythos-5·Fable·12 days after the previous release
A new top tier: a Mythos-class model with safeguards that make it safe for general use. Its abilities topped every model Anthropic had made generally available.
claude-fable-5·Opus·42 days after the previous release
Better coding and knowledge work at the same price as 4.7, a more honest collaborator, and Dynamic Workflows in Claude Code.
claude-opus-4-8·Product
An Anthropic Labs product for making designs, prototypes, slides and one-pagers together with Claude.
·Opus·58 days after the previous release
A stronger Opus with higher resolution vision and a new tokenizer.
claude-opus-4-7·Mythos
Claude Mythos Preview
Restricted accessA frontier model a tier above Opus, shared only with partners in Project Glasswing (AWS, Apple, Google, Microsoft and others) to find and fix security flaws in critical software.
·Sonnet·12 days after the previous release
A Sonnet upgrade at the same $3 / $15 per million token price as Sonnet 4.5.
claude-sonnet-4-6·Opus·73 days after the previous release
The start of the 4.6 generation, the first with dateless model IDs (no more -2025xxxx suffixes).
claude-opus-4-6
2025
7 models·Milestone
Anthropic hands MCP to the new Agentic AI Foundation, co-founded with Block and OpenAI, one year after launching it.
·Opus·40 days after the previous release
A new top model for coding, agents and computer use, priced well below earlier Opus models.
claude-opus-4-5-20251101200K context·Product
Folders of instructions, scripts and resources that Claude loads only when a task needs them. Works across claude.ai, Claude Code and the API.
·Haiku·16 days after the previous release
Sonnet 4 level coding at a third of the cost and more than twice the speed.
claude-haiku-4-5-20251001200K context·Claude Code
Bundles of slash commands, subagents, MCP servers and hooks that install with one command, shared through plugin marketplaces.
·Sonnet·55 days after the previous release
Billed as the best coding model in the world at launch, with big gains on computer use and long agent runs.
claude-sonnet-4-5-20250929200K context·Opus·75 days after the previous release
An incremental Opus upgrade focused on agentic tasks, real-world coding and reasoning.
claude-opus-4-1-20250805200K context·Claude Code
Claude Code leaves preview, with VS Code and JetBrains integrations and background tasks via GitHub Actions.
·Sonnet·87 days after the previous release
Launched with Opus 4 as a large upgrade over 3.7 Sonnet, and made available to free users.
claude-sonnet-4-20250514200K context·Opus·87 days after the previous release
Claude 4 drops the version-first naming. Opus 4 was pitched as the best coding model at launch, able to work for hours on long tasks.
claude-opus-4-20250514200K context·Claude Code
An agentic coding tool that runs in your terminal, released as a limited research preview with Claude 3.7 Sonnet.
·Sonnet·125 days after the previous release
The first hybrid reasoning model: one model that can answer quickly or think step by step with visible extended thinking.
claude-3-7-sonnet-20250219200K context
2024
6 models·Product
An open standard for connecting AI apps to tools and data. It became the common way to plug services into Claude and other assistants.
·Product
Claude can look at a screen, move a cursor, click and type. The first frontier model offered with this in public beta.
·Haiku·124 days after the previous release
Announced with the upgraded 3.5 Sonnet. Surpassed Claude 3 Opus on many benchmarks at Haiku speed.
claude-3-5-haiku-20241022200K context·Sonnet·124 days after the previous release
A big upgrade under the same name, especially for coding. Often called "3.5 Sonnet (new)" or "3.6" by the community.
claude-3-5-sonnet-20241022200K context·Product
A side panel where Claude's code, documents and small apps render live next to the chat.
·Sonnet·99 days after the previous release
A mid-tier model that beat Claude 3 Opus at a fifth of the price. Shipped with Artifacts on claude.ai.
claude-3-5-sonnet-20240620200K context·Haiku·9 days after the previous release
The first Haiku: the fastest and cheapest Claude 3 model, announced with the family and shipped nine days later.
claude-3-haiku-20240307200K context·Sonnet·104 days after the previous release
The first Sonnet: the balanced middle tier of the Claude 3 family, and the model behind free claude.ai at launch.
claude-3-sonnet-20240229200K context·Opus·104 days after the previous release
The first Opus. Top of the new three-tier Claude 3 family (Haiku, Sonnet, Opus) and the first Claude models with vision.
claude-3-opus-20240229200K context
2023
5 models·Claude·104 days after the previous release
Doubled the context window to 200K tokens, cut hallucination rates and added tool use in beta.
claude-2.1200K context·Instant·29 days after the previous release
An update to the fast tier with stronger math, coding and reasoning, built on the strengths of Claude 2.
claude-instant-1.2·Claude·119 days after the previous release
Better coding, math and reasoning, plus longer outputs. Launched alongside claude.ai, the first public chat site for Claude (US and UK).
100K context·Product
Claude's context window grows from 9K to 100K tokens, enough to read a whole novel or a long technical document in one prompt.
100K context·Instant
A lighter, faster and cheaper sibling to Claude, launched the same day. The ancestor of today's Haiku tier.
·Claude
Anthropic's first public model, offered to businesses through an API and partners like Notion, Quora and DuckDuckGo after months of closed testing.
Cite or link this timeline
Free to quote and link. Every entry has its own anchor, so you can point to a single release.
Claude AI Directory, "Claude Release Timeline", updated November 12, 2026. https://www.claudeai.directory/anthropic-claude-release-timelinesSpot a mistake or a missing release?
Email [email protected] with a link to the official source and we will update the page.
Suggest a correctionDates follow Anthropic's announcements and official changelogs. Stats count generally available model releases only; the restricted Mythos models are listed but not counted. Gaps are measured between distinct release days, so models launched together count once. Claude AI Directory is a community site and is not affiliated with Anthropic.