By Sabdar Shah
Claude features 2026, read for government contractors: twelve model launches, skills as shareable folders, and the admin controls that caught up with them. Twelve Claude models launched in 2026 and every one of them will have a successor. The durable changes were the packaging and the admin controls, and this reads the year through them. Claude Features 2026 Ask what Claude features 2026 actually delivered and the honest answer is not a model name. Anthropic shipped twelve named models in twelve months, and every one of them will have a successor. What changed for a government contractor is narrower and more useful. By November an administrator can decide who uses Claude, which systems it reaches and which models they may pick. They also decide whether the platform scans a third-party skill before it runs, and which contract a dollar of spend belongs to. Capability grew all year. The governance surface caught up with it, and that is the half a security questionnaire asks about. Every date below comes from Anthropic's own dated release notes, read on 7 November 2026. ! Three 2026 threads: twelve model launches, skills and plugins as versioned packages, and the governance controls a questionnaire asks about (https://mhhifytmrlyksfrjacvi.supabase.co/storage/v1/object/public/blog-images/2026/11/claude-features-2026-figure.png) The model names churned; the packaging and the controls are what a firm can still build on next year. Twelve Models, One Lesson The Claude apps release notes (https://support.claude.com/en/articles/12138966-release-notes) record a model launch in February, April, May, June, July, September and October. Opus 4.6 on 5 February. Sonnet 4.6 on 17 February, carrying a 1M token context window in beta. Opus 4.7 on 16 April, Opus 4.8 on 28 May, Fable 5 on 9 June, Sonnet 5 on 30 June, Opus 5 on 24 July. Then Fable 5.1 and Mythos 5.1 together on 1 September, Opus 5.5 on 22 September, Sonnet 5.5 on 28 September and Haiku 5.5 on 7 October. Twelve launches in one year. Pin a documented process to a model name and it breaks within a quarter. One entry makes the point harder than the count does. On 12 June Anthropic suspended access to Fable 5 and Mythos 5, and restored it on 1 July. A model your team relies on can disappear for three weeks. Write your runbook around the steps and the review gates, then let the model underneath change. Skills Became Shareable Assets The single most useful shift for a proposal shop started in December 2025. Skills gained organization-wide management for Team and Enterprise plans, a directory of partner-built skills, and an open standard. Agent Skills (https://agentskills.io/home) describes that standard plainly. Skills are "a lightweight, open format for extending AI agent capabilities with specialized knowledge and workflows", and "a skill is a folder
Twelve Claude models launched in 2026 and every one of them will have a successor. The durable changes were the packaging and the admin controls, and this reads the year through them.
Claude Features 2026 Ask what Claude features 2026 actually delivered and the honest answer is not a model name. Anthropic shipped twelve named models in twelve months, and every one of them will have a successor. What changed for a government contractor is narrower and more useful. By November an administrator can decide who uses Claude, which systems it reaches and which models they may pick. They also decide whether the platform scans a third-party skill before it runs, and which contract a dollar of spend belongs to. Capability grew all year. The governance surface caught up with it, and that is the half a security questionnaire asks about. Every date below comes from Anthropic's own dated release notes, read on 7 November 2026. ! Three 2026 threads: twelve model launches, skills and plugins as versioned packages, and the governance controls a questionnaire asks about (https://mhhifytmrlyksfrjacvi.supabase.co/storage/v1/object/public/blog-images/2026/11/claude-features-2026-figure.png) The model names churned; the packaging and the controls are what a firm can still build on next year. Twelve Models, One Lesson The Claude apps release notes (https://support.claude.com/en/articles/12138966-release-notes) record a model launch in February, April, May, June, July, September and October. Opus 4.6 on 5 February. Sonnet 4.6 on 17 February, carrying a 1M token context window in beta. Opus 4.7 on 16 April, Opus 4.8 on 28 May, Fable 5 on 9 June, Sonnet 5 on 30 June, Opus 5 on 24 July. Then Fable 5.1 and Mythos 5.1 together on 1 September, Opus 5.5 on 22 September, Sonnet 5.5 on 28 September and Haiku 5.5 on 7 October. Twelve launches in one year. Pin a documented process to a model name and it breaks within a quarter. One entry makes the point harder than the count does. On 12 June Anthropic suspended access to Fable 5 and Mythos 5, and restored it on 1 July. A model your team relies on can disappear for three weeks. Write your runbook around the steps and the review gates, then let the model underneath change. Skills Became Shareable Assets The single most useful shift for a proposal shop started in December 2025. Skills gained organization-wide management for Team and Enterprise plans, a directory of partner-built skills, and an open standard. Agent Skills (https://agentskills.io/home) describes that standard plainly. Skills are "a lightweight, open format for extending AI agent capabilities with specialized knowledge and workflows", and "a skill is a folder