Claude Haiku 4.5
Claude Haiku 4.5, released October 15, 2025, is Anthropic's fastest model delivering near-frontier performance at blazing speed with breakthrough efficiency. It runs 2x faster than Sonnet 4 while achieving 90% of Sonnet 4.5's agentic coding capabilities on complex software engineering tasks, making it ideal for real-time applications and high-volume workloads requiring both exceptional speed and advanced intelligence.
Core Capabilities
- 2x faster than Sonnet 4: Processes tasks at twice the speed of the previous generation, delivering rapid response times that enable seamless real-time interactions and instant feedback loops in production applications
- 90% agentic performance: Achieves 90% of
Sonnet 4.5's performance in agentic coding evaluations, demonstrating near-frontier capabilities in autonomous code generation, debugging, and complex problem-solving workflows - Similar to Sonnet 4: Provides coding quality comparable to
Claude Sonnet 4, maintaining high standards for syntax accuracy, architectural decisions, and code comprehension across multiple programming languages - Enhanced safety: Operates at
AI Safety Level 2 (ASL-2)with improved alignment overHaiku 3.5, offering better resistance to prompt injection attacks and more reliable adherence to safety guidelines - Cost efficient:
$1/$5pricing delivers exceptional value per token at high speed, offering 3x cost savings compared toSonnet 4.5($3/$15) while maintaining near-frontier performance levels - Multi-agent orchestration: Can be coordinated by
Sonnet 4.5for complex parallel task execution, enabling efficient distribution of work across multiple specialized agents in sophisticated multi-agent systems
Technical Specifications
- Model ID:
claude-haiku-4-5-20251015 - Pricing:
$1/$5per million tokens (input/output) - Availability: Claude Code, Claude apps, Claude API, Amazon Bedrock, Google Cloud Vertex AI
- Release Date: October 15, 2025
Best Used For
- Chat assistants: Real-time conversational interfaces requiring instant responses, where sub-second latency is critical for maintaining natural dialogue flow and user engagement in customer-facing applications
- Customer service: High-volume support automation with quick turnaround times, enabling businesses to handle thousands of concurrent support tickets while maintaining intelligent, context-aware responses at minimal cost
- Pair programming: Rapid prototyping and code generation in development workflows, where developers need immediate intelligent suggestions, refactoring assistance, and quick iterations during active coding sessions
- Computer use tasks: Fast agent operations interacting with interfaces and web browsers, where speed is essential for smooth autonomous navigation, form filling, and UI interaction workflows
- Multi-agent systems: Efficient parallel execution when orchestrated by
Sonnet 4.5, allowing lightweight worker agents to handle specific subtasks at high speed while the orchestrator coordinates complex multi-step workflows
Usage
# Set as default model
export ANTHROPIC_MODEL="claude-haiku-4-5-20251015"
# Use for specific task
claude --model claude-haiku-4-5-20251015 "Your task here"
# Or use /model command in Claude Code
/model
See Also: Model Comparison|Sonnet 4.5|Change Model|Pricing

