May 29, 01:03 AM

OpenAI's GPT-2 (124M) Now Replicable in 90 Minutes for $20; GPT-5 Training Costs Estimated at $1.7-2.5 Billion. New Llama3-V SOTA Model Outperforms Previous VLMs.

OpenAI's GPT-2 (124M) model, released in 2019, has become more accessible due to reduced compute costs. It can now be replicated in 90 minutes for $20 using llm.c. The cost to train GPT-5 is estimated to range from $1.7 to $2.5 billion, significantly more expensive than its predecessors. Additionally, a new SOTA open-source VLM model, Llama3-V, has been introduced, outperforming previous models like LLaVA and offering comparable performance to GPT4-V and Gemini Ultra with a significantly smaller model size.

#OpenAI #GPT #LLaVA #Gemini Ultra

Written with ChatGPT (GPT-3).

Sources

Sambhav Gupta@sambhavgupta6
2 years ago
K2-65B🏔️, the most performant fully-open LLM released to date. As a blueprint for open-source AGI, they release all model checkpoints, code, logs, and data. About K2: 🧠65 billion parameters 🪟Fully transparent & reproducible 🔓Apache 2.0 📈Outperforms Llama 2 70B https://t.co/2XGw7YoncF
Hyperbolic@hyperbolic_labs
2 years ago
Our cofounder @Yuchenj_UW managed to train GPT-2 using @karpathy’s llm.c framework in just 27 minutes for under $10. Kudos to @karpathy’s contribution to open source AI. The future of AI is collaborative 🤘 https://t.co/rsKf872qM2
Rohan Paul@rohanpaul_ai
2 years ago
Llama 3-V: Close to matching GPT4-V with a 100x smaller model and 500 dollars https://t.co/9o7uneOphM

Additional media

Image #1 for story openai-s-gpt-2-124m-now-replicable-90-minutes-20-gpt-5-costs