What is Claude?
Claude is the artificial intelligence assistant developed by Anthropic, founded in 2021 by former OpenAI members, including Dario and Daniela Amodei. Claude stands out for its focus on AI Safety and its extended reasoning capabilities.
In November 2025, Claude Code reached $1 billion in annualized revenue in just 6 months since public launch, establishing Claude as the leader in AI-powered coding tools.
Model Family (December 2025)
Claude Opus 4.5 (November 2025)
- Anthropic's most intelligent model
- State-of-the-art in software engineering and agentic tasks
- New
effort parameter to control reasoning depth
- Reduced pricing: $5/$25 per million tokens (previously $15/$75)
- Context window: 200K tokens
- Max output: 64K tokens
- Knowledge cutoff: March 2025
Claude Sonnet 4.5 (September 2025)
- "The best coding model in the world" according to Anthropic
- "The best model for agents"
- 1M token context window (beta)
- Extended thinking with granular control
- Native context awareness
- Pricing: $3/$15 per million tokens
Claude Haiku 4.5 (October 2025)
- Near-frontier performance at maximum speed
- First Haiku with extended thinking and computer use
- Ideal for multi-agent (orchestrated by Sonnet/Opus)
- Pricing: $1/$5 per million tokens
- Safest model per internal evaluations (ASL-2)
Main Features
Reasoning Capabilities
- Extended Thinking: Visible and controllable step-by-step reasoning
- Effort Parameter (Opus 4.5): Depth control (low/medium/high)
- Thought Signatures: Maintains reasoning chain in multi-turn flows
- Context Awareness: Real-time tracking of available context
Tools and Functionality
- Claude Code: Agentic coding tool in terminal
- Computer Use: Computer control for automation
- Tool Search: Dynamic search among hundreds of tools
- Agent Skills: Extensions for specialized tasks (Excel, Word, PowerPoint, PDF)
- Web Fetch: Web content retrieval
- Citations: Source attribution in responses
- MCP (Model Context Protocol): Protocol for external tool integration
Integrations
- GitHub Copilot
- Cursor
- VS Code
- Microsoft Excel (Claude for Excel)
- Google Chrome (Claude for Chrome)
- Slack
- Notion
- AWS Bedrock
- Google Vertex AI
- Microsoft Foundry
Plans and Pricing (December 2025)
Consumer Plans
| Plan | Price | Main Features |
|---|
| Free | $0 | Limited access, ~20 searches/day |
| Pro | $20/month | 5x more usage than Free, all models |
| Max Expanded | $100/month | 5x more usage than Pro, priority access |
| Max | $200/month | 20x more usage than Pro, experimental features |
| Team | $25-30/user/month | Shared workspace, admin controls |
| Enterprise | ~$60/user/month | SSO/SCIM, audit logs, dedicated support |
API Pricing
| Model | Input | Output | Thinking |
|---|
| Opus 4.5 | $5/MTok | $25/MTok | - |
| Opus 4.1 | $15/MTok | $75/MTok | $30/MTok |
| Sonnet 4.5 | $3/MTok | $15/MTok | $6/MTok |
| Haiku 4.5 | $1/MTok | $5/MTok | $2/MTok |
Available Discounts
- Batch API: 50% discount on asynchronous processing
- Prompt Caching: Up to 90% savings on cached tokens
- Nonprofits: Up to 75% discount on Team/Enterprise plans
Claude Code
Claude Code is the agentic coding tool that lives in your terminal:
- Public launch: May 2025
- $1B annualized revenue: November 2025
- Customers: Netflix, Spotify, KPMG, L'Oréal, Salesforce
- Bun acquisition: December 2025 (JavaScript runtime)
- SDK available for TypeScript, Python, CLI
- Open source on GitHub: github.com/anthropics/claude-code
Security and Privacy
Anthropic is a leader in AI safety with its Responsible Scaling Policy:
- ASL-2: Haiku 4.5 (limited risk)
- ASL-3: Sonnet 4.5, Opus 4.5, Opus 4.1 (advanced controls)
- Opus 4.5: Most resistant to prompt injection of any frontier model
- Encryption in transit and at rest
- SOC 2 Type 2 compliance
- Option to not train on user data
Technical Specifications
| Model | Context | Max Output | Knowledge Cutoff |
|---|
| Opus 4.5 | 200K | 64K | March 2025 |
| Sonnet 4.5 | 200K (1M beta) | 64K | January 2025 |
| Haiku 4.5 | 200K | 64K | February 2025 |
