HAVE AI NEWS HAVE AI NEWS
tg subscribe
Models

Claude Sonnet 5.5: Anthropic's Mid-Tier Model Delivers Near-Opus Performance at Half the Price

Claude Sonnet 5.5: Anthropic's Mid-Tier Model Delivers Near-Opus Performance at Half the Price

Anthropic launches Claude Sonnet 5.5, a mid-tier model that nearly matches the capabilities of its flagship Opus 5.5 at lower cost and faster speed. It excels in knowledge work and coding while introducing new security features. Includes a simple test to select the right Claude model and coverage of AI safety concerns plus a major hardware deal.

Claude Sonnet 5.5: Anthropic's Mid-Tier Model Delivers Near-Opus Performance at Half the Price

Anthropic has released Claude Sonnet 5.5, a mid-tier model in the Claude 5.5 family that performs close to the company's top-tier Opus 5.5 model while offering substantial cost savings.

The new model delivers 30% faster processing speeds and notable improvements over the previous Sonnet version in knowledge work and coding tasks. It scores 56 on the Artificial Analysis Intelligence Index, placing it just behind Opus 5.5 and ahead of Fable 5.1.

Sonnet 5.5 nearly ties Opus 5.5 on several office productivity tests and ranks at or near the level of both Opus 5.5 and other leading models on development benchmarks. It also features the first Sonnet-tier cyber security guardrails, matching the security restrictions previously seen in higher-tier models.

Pricing and Efficiency Gains

  • Jobs cost up to 30% less than with the prior Sonnet version.
  • Capabilities for low and medium effort tasks are available at around one-tenth the cost of the previous top model.

A Quick Test for Selecting the Right Claude Model

To identify which Claude model best fits your needs, run a straightforward comparison test. Open separate chats and use identical settings for Opus 5.5, Opus 5, and Fable 5.1. Paste the same fictional brief containing two hidden errors. Evaluate each model's accuracy in spotting the errors, the effort required to produce corrected outputs, and the final quality.

The test recommends defaulting to Sonnet 5.5 for most cases due to its balanced performance. It also recorded an API cost of $0.45 for Fable 5.1 compared to $0.18 for Sonnet 5.5.

Other Notable AI Developments

Leading AI researchers, including those from Anthropic and OpenAI, along with pioneers Geoffrey Hinton and Yoshua Bengio, co-authored a paper on preparing for an intelligence explosion from self-improving AI systems. The report outlines scenarios where progress could accelerate sharply, such as reducing a year's advancements to five weeks. It also notes Anthropic's internal tracking shows AI now completing 26% of the lab's R&D work independently, up from 1% earlier this year.

AMD has agreed to acquire Fei-Fei Li's World Labs in an $8.2 billion all-stock deal. Li, the founder of the 3D world generation startup, will join AMD as chief scientist to integrate these models more closely with hardware, building on ongoing collaboration since last year.

Author: Zach Mink

Source: www.therundown.ai