Claude Sonnet 5
AnthropicLarge Language ModelProprietaryAnthropic's best combination of speed and intelligence. 1M token context with adaptive thinking and fast response times, at $3 per million input tokens (introductory $2 through Aug 31, 2026).
Abilities
Use Cases
Available in Tools
Availability
How to Use
Pros
- Excellent speed-intelligence trade-off — fast responses with near-Opus-4.8 quality
- 1M token context window — same massive context as Opus 4.8 at a lower price
- Adaptive thinking support — reasons more deeply on hard problems when needed
- Introductory pricing of $2/$10 per MTok through Aug 31 2026 — significantly below list price
- 128K max output — generates large files and long-form content without truncation
Cons
- Quality gap vs Opus 4.8 on the most complex reasoning tasks
- Closed source — no self-hosting
- Introductory pricing expires Aug 31 2026 — list price rises to $3/$15 per MTok
What to Use It For
Perfect For
Fast enough for interactive use while delivering near-Opus quality for complex refactoring
1M context window handles long documents at a price point viable for high-volume enterprise use
Optimal balance of speed, quality, and cost for iterative development workflows
Good For
Fast, high-quality prose for blogs, reports, and marketing copy
Reliable structured output at reasonable cost for ETL and document parsing
Not Recommended
Drops nuance on highly specialized academic analysis — Opus 4.8 handles this better
Try instead: Claude Opus 4.8, Claude Fable 5
Cloud-only model — requires Anthropic API
Try instead: Gemma 3, Phi-4
Do Not Use For
Text-only LLM — no image creation capabilities
Try instead: DALL-E 3, FLUX.1
At $3/1M input Haiku 4.5 at $1/1M is 3x cheaper for bulk classification
Try instead: Claude Haiku 4.5