Anthropic currently offers four Claude models in active use, each with its own balance of intelligence, speed, and price. Here's an up-to-date comparison.
Comparing Claude models
Haiku 4.5
Fast
Simple, high-volume tasks
Price (input)$1 / MTok
Price (output)$5 / MTok
Context200K tokens
ThinkingNo
SpeedVery fast
Best forHigh-volume tasks
Sonnet 4.6
Standard
Predecessor to Sonnet 5
Price (input)$3 / MTok
Price (output)$15 / MTok
Context1M tokens
ThinkingYes
SpeedMedium
Best forEveryday tasks
Recommended
Sonnet 5
Balanced
Best price/performance
Price (input)$2 / MTok
Price (output)$10 / MTok
Context1M tokens
ThinkingYes (adaptive)
SpeedFast
Best forMost tasks
Opus 5
Powerful
Complex, long-form tasks
Price (input)$5 / MTok
Price (output)$25 / MTok
Context1M tokens
ThinkingYes (adaptive)
SpeedSlower
Best forVery complex analysis
Fable 5
Top model
Requires Max plan / credits
Price (input)$10 / MTok
Price (output)$50 / MTok
Context1M tokens
ThinkingAlways on (adaptive)
SpeedSlowest
Best forThe hardest tasks
Benchmarks (SWE-bench Verified)
Indicative figures, not officially confirmed by Anthropic
Fable 5~95%
Opus 5~89%
Sonnet 5~85%
Sonnet 4.6~60%
Recommended use
Complex analysis
Opus 5
Opus 5
Everyday tasks
Sonnet 5
Sonnet 5
Fast answers
Haiku 4.5
Haiku 4.5
The hardest tasks
Fable 5 (Max plan)
Fable 5 (Max plan)
Quick reference: best for and speed
Model
Best for
Speed
Fable 5
Absolute top model, requires Max plan
Slower
Opus 5
Very complex analysis and long documents
Slow
Sonnet 5
Strong balance between intelligence and speed
Medium
Haiku 4.5
Fast, simple answers, cheapest
Fast
Sonnet 4.6
Previous model — everyday tasks
Medium
Sonnet 5 usually hits the best balance of intelligence and speed for everyday tasks, while Opus 5 and Fable 5 are worth saving for the hardest problems.