The state of Claude: models, capabilities and access
All of Claude on one page, kept current: which models exist today, what they can do, and where to use them. Snapshot current as of June 2026; exact figures (pricing, limits, context sizes) change fast and should be checked on anthropic.com.
Latest Claude & Anthropic news
- Claude Fable 5 by Anthropic, releasing tomorrow — Hacker News
- Ask HN: Is Claude Fable 5 built from scratch or just better data? — Hacker News
- Tell HN: Anthropic's Fable model is too expensive — Hacker News
- Show HN: I used Claude Mythos to build my startup in 1 day — Hacker News
- Vous ne pouvez plus utiliser Claude Fable 5, l’IA ultra-puissante d’Anthropic : l’administration Trump bloque l’accès de “tout ressortissant étranger” à ce modèle — Journal du Geek
- Anthropic présente ses excuses pour les mesures de sécurité invisibles de Claude Fable 5. L'entreprise altère délibérément les réponses aux prompts soupçonnés de viser la distillation du modèle — Developpez.com
- Anthropic dévoile Claude Fable 5, premier modèle public de la classe Mythos — blogdumoderateur.com
- Anthropic dévoile Claude Fable 5, son modèle d’IA le plus performant à ce jour — TradingView
- Claude Fable 5 : le modèle "Mythos" qu'Anthropic jugeait trop dangereux est maintenant entre vos mains — Les Numériques
- Claude: Elevated Error Rates for Opus 4.8, Opus 4.7, Opus 4.6, and Sonnet 4.6 — Hacker News
Claude's models today
The line-up trades off capability, speed and cost. Opus 4.8 (API id claude-opus-4-8) is the most capable model, for hard reasoning and complex code. Sonnet 4.6 (claude-sonnet-4-6) is the balanced default for most tasks. Haiku 4.5 (claude-haiku-4-5) favors speed and cost, for volume and real time. Fable 5 (claude-fable-5) is a newer family. Exact pricing, rate limits and context sizes change often: refer to Anthropic's official page.
What Claude can do
Beyond text, Claude reads images (vision) and reasons over long documents. It calls tools and functions, can operate a computer (computer use), and connects to your data and apps through MCP, the Model Context Protocol, an open integration standard. For development, Claude Code acts as an agentic coding assistant, and Artifacts let you generate and then iterate on code or documents in a dedicated panel. Note: Claude does not generate images, it analyzes them.
Where to use Claude
Claude is available on claude.ai (web, macOS and Windows desktop apps, mobile), via Anthropic's API (the Messages API, to build Claude into your own products) and via Claude Code in the terminal, IDE or browser. Pick by use: the interface for daily work, the API to automate, Claude Code to build.
Safety, publisher and independence
Claude is built by Anthropic, a safety-focused AI company, following what it calls Constitutional AI, where explicit principles guide the model's behavior. Claude News is an independent outlet covering Claude and Anthropic: we are not affiliated with Anthropic, and both Claude and Anthropic belong to their respective owners.
Frequently asked questions
What are the current Claude models?
Opus 4.8 (most capable), Sonnet 4.6 (balanced, default), Haiku 4.5 (fast and economical) and Fable 5 (a newer family). For current pricing and limits, see anthropic.com.
Which Claude model should I choose?
Opus for hard reasoning and complex code; Sonnet as the versatile default; Haiku for speed and high volume.
What are MCP and Claude Code?
MCP (Model Context Protocol) is an open standard for connecting Claude to tools and data. Claude Code is Anthropic's agentic coding assistant, usable in the terminal, IDE or browser.
Claude News is published by Héra SASU. Independent media, not affiliated with Anthropic.