• AI Box
  • Posts
  • 🤖 Anthropic a $4.1B OpenAI Rival Launches "Claude 2" Beating ChatGPT-4

🤖 Anthropic a $4.1B OpenAI Rival Launches "Claude 2" Beating ChatGPT-4

🔍 GPT-4 Leak: Massive Size and Parameters. 🧠 AMD Says: AI to Dominate Chip Design

Author

Jaeden Schafer
July 12, 2023

Welcome to AI Box, your weekly source for the latest developments in AI technology and applications.

Today's AI News lineup: 🥊

  • 🤖 Claude 2 Launches: Challenging GPT-4’s Dominance

  • 🔍 GPT-4 Leak: Massive Size and Parameters

  • 🧠 AMD Says: AI to Dominate Chip Design

🤖 Anthropic a $4.1B OpenAI Rival Launches "Claude 2" Beating ChatGPT-4

It can summarize up to 75,000 words, a significant improvement over ChatGPT's 3,000-word limit.

With a knowledge cutoff of early 2023, Claude-2 surpasses GPT-4's capabilities on various tasks, including GRE writing, HumanEval coding, and the Bar exam.

It also possesses features akin to ChatGPT Code Interpreter, enabling it to import and summarize entire PDF documents, all while being 5X more affordable for developers compared to GPT-4.

🔍 GPT-4's Details are Leaked

💡 GPT-4 is revealed to be over 10 times the size of GPT-3, with approximately 1.8 trillion parameters across 120 layers. OpenAI utilized a mixture of experts (MoE) model with 16 experts, each containing around 111 billion parameters. The model is trained on approximately 13 trillion tokens, including millions of rows of instruction fine-tuning data.

💰 The training cost for GPT-4 is estimated to be around $63 million, with extremely low utilization due to numerous failures requiring restarts. Inference costs for GPT-4 are approximately 3 times higher than the 175 billion parameter Davinci model. The model also includes a separate vision encoder for multi-modal capabilities, fine-tuned with an additional 2 trillion tokens.

🧠💻 AMD: AI to Dominate Chip Design

AMD's chief executive, Lisa Su, believes that artificial intelligence (AI) will eventually dominate certain areas of chip design due to the increasing complexity of modern processors.

The company is already utilizing AI in semiconductor design, testing, and verification, with plans to leverage generative AI in future chip design applications.

AI is being employed in chip design for better performance and energy efficiency, as well as in verification suites to reduce bug detection time. Synopsys, an electronic design automation (EDA) tool maker, is ahead in offering AI-enabled software for chip development.

💻 AI Box (Sponsor)

AI Box is a no-code, AI app building platform paired with the App Store for AI that lets you monetize your AI tools.

The platform lets you build apps by linking together hundreds of AI models like ChatGPT, MidJourney, and Eleven Labs. Eventually we’ll integrate software like Gmail, Trello and Salesforce so you can use AI to automate every function within your organization.

To get notified when we launch and be the first to build on the platform just stay subscribed to this newsletter alternatively you can join the waitlist at AIBox.ai

Interested in sponsoring The AI Box Newsletter?
Ask us about our sponsorship options to see if you'd be a good fit!

#️⃣ AI Twitter

Elon dives into AI (again)…

That's all for today!

If you have any interesting projects or ideas, please reach out to us by responding to this email or by sending us a DM on Twitter: @jaeden_ai & @aibox_ai

As always, thanks for reading, and see you next time. 🫡