Claude 3.5 Sonnet and Haiku got an upgrade Now it can even use your computer😯

Anthropic Unveils Upgraded AI Models and Groundbreaking Computer Use Capability

Anthropic, a leading artificial intelligence research company, has announced significant advancements in their AI technology. The company has introduced an upgraded version of Claude 3.5 Sonnet, a new model called Claude 3.5 Haiku, and a revolutionary computer use capability. These developments mark a substantial leap forward in AI capabilities, particularly in the realms of coding, problem-solving, and human-like computer interaction.

Upgraded Claude 3.5 Sonnet: A Coding Powerhouse

The upgraded Claude 3.5 Sonnet model demonstrates impressive improvements across various benchmarks, with particularly notable gains in coding and tool use tasks. Here's a breakdown of its key enhancements:

1. Coding Performance:

   • Improved performance on SWE-bench Verified from 33.4% to 49.0%
   • Outperforms all publicly available models, including specialized systems for agentic coding

2. Tool Use Capability:

   • Enhanced performance on TAU-bench:
     • Retail domain: Improved from 62.6% to 69.2%
     • Airline domain: Increased from 36.0% to 46.0%

3. Cost and Speed:

• Offers these advancements at the same price and speed as its predecessor

Real-World Impact

Early feedback from industry partners highlights the practical benefits of the upgraded Claude 3.5 Sonnet:

• GitLab: Reported up to 10% stronger reasoning across use cases with no added latency
• Cognition: Experienced substantial improvements in coding, planning, and problem-solving
• The Browser Company: Noted that Claude 3.5 Sonnet outperformed every model they've tested before

These improvements make Claude 3.5 Sonnet an ideal choice for powering multi-step software development processes and automating web-based workflows.

Introducing Claude 3.5 Haiku: Power Meets Efficiency

Anthropic is also launching Claude 3.5 Haiku, the next generation of their fastest model. This new model offers:

1. Improved Performance:

   • Surpasses Claude 3 Opus (Anthropic's previous largest model) on many intelligence benchmarks
   • Particularly strong in coding tasks, scoring 40.6% on SWE-bench Verified

2. Cost-Effective Solution:

   • Delivers enhanced capabilities at the same cost and similar speed to Claude 3 Haiku

3. Specialized Capabilities:

• Low latency
• Improved instruction following
• More accurate tool use

Claude 3.5 Haiku is well-suited for user-facing products, specialized sub-agent tasks, and generating personalized experiences from large datasets.

Groundbreaking Computer Use Capability

Perhaps the most exciting announcement is the introduction of computer use capability, now available in public beta. This innovative feature allows Claude to interact with computers in a human-like manner, opening up new possibilities for automation and task completion.

Key Features of Computer Use:

1. Human-like Interaction: Claude can look at screens, move cursors, click buttons, and type text.

2. Wide-ranging Applications:

   • Automating repetitive processes
   • Building and testing software
   • Conducting open-ended research tasks

3. Performance: On OSWorld, which evaluates AI models' ability to use computers like humans, Claude 3.5 Sonnet scored:

• 14.9% in the screenshot-only category (compared to the next-best AI system's score of 7.8%)
• 22.0% when given more steps to complete tasks

Early Adopters and Use Cases

Several prominent companies have already begun exploring the possibilities of computer use:

• Asana
• Canva
• Cognition
• DoorDash
• Replit: Using Claude 3.5 Sonnet's capabilities for UI navigation in their Replit Agent product
• The Browser Company

These companies are leveraging Claude's ability to perform tasks requiring dozens, and sometimes hundreds, of steps to complete.

Responsible Development and Deployment

Anthropic emphasizes its commitment to responsible AI development:

1. Safety Testing: Joint pre-deployment testing conducted with the US AI Safety Institute (US AISI) and the UK Safety Institute (UK AISI)

2. Risk Assessment: Evaluated for catastrophic risks, adhering to the ASL-2 Standard outlined in Anthropic's Responsible Scaling Policy

3. Proactive Safety Measures: Developed new classifiers to identify computer use and potential harm

Looking Ahead

As these new models and capabilities are deployed, Anthropic anticipates rapid improvements and refinements. The company encourages developers to explore these new tools, particularly the computer use beta, while being mindful of its current limitations and potential risks.

Conclusion

Anthropic's latest announcements represent a significant leap forward in AI technology. The upgraded Claude 3.5 Sonnet, the new Claude 3.5 Haiku, and the groundbreaking computer use capability offer exciting possibilities for developers, businesses, and researchers. As these tools continue to evolve, they have the potential to revolutionize how we interact with computers and automate complex tasks. However, it's crucial to approach these advancements responsibly, considering both their immense potential and possible implications for society.