Then why is OpenAI releasing GPT-4.5 as the last non-reasoning model? They would need to scale up all the way to GPT-AGI.
@arrakis_ai
-
OpenAI Orion Project Abandoned for Inference Models Strategy
By
–
The only thing I can hope for is that OpenAI abandoned the Orion project in November 2023 and redirected all their efforts into inference models. It feels like they deliberately released this expensive, underperforming model just to tell us to chill and cut our expectations by
-
OpenAI Deep Research Usage Limits Communication Transparency
By
–
Did OpenAI originally inform the remaining number of Deep Research uses?
-
Grok 3 Voice Mode Offers Uncensored AI Conversation Features
By
–
Grok 3’s voice mode has no censorship. It’s quite surprising.
— CHOI (@arrakis_ai) 21 février 2025
Grok Voice Chat with ChatGPT pic.twitter.com/wqxskfpoWYGrok 3’s voice mode has no censorship. It’s quite surprising. Grok Voice Chat with ChatGPT
-
Grok’s H100 GPU Training: Missing Transparency on Key Metrics
By
–
The number of H100 GPUs used for training helps estimate parameters, data size, and training duration, so it’s an important detail. But Grok’s team hasn’t been clear about key factors like benchmark performance and the exact number of GPUs used for training, which leads to
-
Grok-3 Training Hardware: 100k vs 200k H100s Clarification
By
–
I want to clarify this. Was Grok-3 trained on 200k H100s, or is it being served using them? I remember hearing in the live Q&A that it was trained on 100k H100s.
-
Grok 3 Image Generation Updates Compared to Grok 2
By
–
Has image generation in Grok 3 been updated from image generation in Grok2?
-
AI Model Codeforces Performance Doubles in Two Years
By
–
two years ago, we were excited to see a model with a Codeforces Elo of 392.
-
Pro Plan Deep Research Limitation Frustrates User Subscription
By
–
If I had known that I couldn't use 25 'Deep Research' per day, I wouldn't have subscribed to the Pro plan.
