Anthropic launches Claude Sonnet 5.5, touting faster performance and lower costs
Anthropic officially launched Claude Sonnet 5.5, its latest mid-tier AI model, following earlier leaks that hinted at its imminent release and capabilities. The company positions Sonnet 5.5 as a significantly faster and more cost-efficient solution for routine tasks, aiming to be a practical work partner. This release marks Anthropic's second model launch within a short period, expanding its lineup with a focus on accessibility and performance. Anthropic claims Sonnet 5.5 offers performance nearing that of its flagship Opus 5.5 model, particularly excelling in coding tasks, while maintaining the same API pricing as its predecessor, Sonnet 5. Many outlets report Anthropic's claims of a 30% speed increase and up to 30% lower costs per task. The model also achieved a 70.6% score on Terminal-Bench 4.0, with some outlets stating it surpasses Opus 5.5 on this metric. It also incorporates a "safety router" for enhanced security.
86 articles from 77 outlets covered this story. Their coverage differs on 2 points. The underlying claim is sourced from a shipped.
What do all outlets agree on?
77 outlets covered “Anthropic launches Claude Sonnet 5.5, touting faster performance and lower costs”. All of them report the following:
- Anthropic launched Claude Sonnet 5.5
- It is a mid-tier AI model
- Anthropic claims it is faster than its predecessor
- Anthropic claims it is more cost-efficient per task
- It maintains the same API pricing as Sonnet 5
- It is available on platforms like AWS and Snowflake
- It includes a "safety router"
- It was released shortly after another model
- It scored 70.6% on Terminal-Bench 4.0
Did outlets disagree about this?
Yes. Coverage of “Anthropic launches Claude Sonnet 5.5, touting faster performance and lower costs” differs on 2 points. Each account below is how a different outlet described the same event:
Claude Sonnet 5.5's performance relative to Opus 5.5
The Decoder, thenewstack.io, The Rundown AI, Yahoo Tech (nearly matches Opus 5.5 on benchmarks); Decrypt, Yahoo Tech, Dealroom.co, Android Headlines (beats Opus 5.5 at coding); mixed-news.com (scores above Opus 5.5 on Terminal-Bench)
Claude Sonnet 5.5's cost reduction per task compared to Opus 5.5
The Decoder, TradingKey, Digital Trends, NDTV Profit, analyticsindiamag.com, Yellow.com (up to 30% cheaper per task); thenewstack.io, Decrypt, Yahoo Tech, 디지털투데이, eu.36kr.com, mixed-news.com, Android Headlines (half the price)
Which figures do outlets report differently?
3 figures in this story are reported with conflicting values:
Speed increase over previous generation30% faster vs over 30% faster
Cost reduction per task compared to Opus 5.5up to 30% cheaper vs half the price
API pricing$2 Per Million Input Tokens vs unchanged Sonnet 5 pricing vs same price vs $2/$10 Price
Which outlets covered this?
All 86 articles found on this story, grouped by the stance of the piece. Every link goes to the original publisher.
optimist13
pragmatist73
Introducing Claude Sonnet 5.5 on AWSoriginalAWS Machine Learning Blog · company statement