OpenAI is actively preparing for the GPT-5 launch. New assets have been added for GPT-5, GPT-5 Mini and GPT-5 Nano, signalling that these models will become available in the API soon. Soon h/t @legit_api
COMPUTING
-
Model Communication Protocol: From Text-Based Design to Dominance
By
–
Model Communication is going to happen slowly, and then all at once:
— dr. jack morris (@jxmnop) 6 août 2025
Level 0: We design a text-based protocol for models & programs to communicate (read: MCP)
Level 1: Text-based model communication grows to exceed human communication in worldwide internet bitstream volume… pic.twitter.com/3vP6yPBmpTModel Communication is going to happen slowly, and then all at once: Level 0: We design a text-based protocol for models & programs to communicate (read: MCP)
Level 1: Text-based model communication grows to exceed human communication in worldwide internet bitstream volume -
Secure Lightweight AI Platform for Regulated Industries
By
–
North has security and data privacy at its core, designed for regulated industries and government agencies. Its lightweight architecture is optimized for private deployments with as few as two GPUs for performance on-prem or in hybrid clouds, VPCs, or airgapped environments.
-
Axelera Engineers Power Metis AI Edge Platform Innovation
By
–
On #ProfessionalEngineersDay, we salute the brilliant minds behind the technologies shaping tomorrow – especially our own Axelera engineers! From silicon to software, your work powers the Metis AI platform and redefines what’s possible at the edge. Thank you team! #EdgeAI
-
PyTorch Team Shows Efficient Llama Training Methods
By
–
Yes I know. We helped show the PyTorch team how to train Llama efficiently – they have a whole GPU Mode talk about it. These issues do not at all line with your claims however.
-
Software Eating World While AI Digests Technology
By
–
Software is eating the world, and AI is digesting it.
-
GLM 4.5 Air outperforms gpt-oss-20b despite higher RAM usage
By
–
Not sure yet! GLM 4.5 Air on my laptop impressed me more than gpt-oss-20b but also used nearly 4x the RAM
-
LM Studio and Groq/Cerebras for Large Language Model Inference
By
–
LM studio for 20B model and Groq/Cerebras for 120B model.
-
H100 GPU Pricing Estimates and Model Card Analysis
By
–
I got those H100 pricing estimates from here: https://
thundercompute.com/blog/nvidia-h1
00-pricing
… More of my notes on the model card: -
H100 GPU costs for 120B and 20B LLM model training
By
–
According to the model card gpt-oss-120b took 2.1 million H100-hours and the 20b model 1/10th of that I found quotes for H100 pricing ranging from $2/hour to $11/hour, which would indicate that the 120b model cost between $4.2m and $23.1m and the 20b between $420,000 and $2.3m