Comparing & Contrasting Recent LLMs Architecture > DeepSeek-V3/R1
> OLMo 2
> Gemma 3
> Mistral Small 3.1
> Llama 4
> Qwen3 (dense+MoE)
> SmolLM3
> Kimi 2
> GPT-OSS Are 2025 LLMs really that different from each other? MoE, MLA, GQA, sliding window, normalization games & more.
@theahmadosman
-
2025 LLMs Compared: Architecture Differences and Technical Innovations
By
–
-
Google’s Ambitious World Modeling Project This Century
By
–
Google despite whatever we say about it fumbling and what not has been the most ambitious project in this century they've the world perfectly modeled from every aspect and depth
-
240V 40A Breaker Setup for Multi-GPU Mining Configuration
By
–
40amp 240v breaker is more than enough easy to add in any house built within the last 15 years then you daisy chain 3x 2000w Platinum 95+ PSU more in the Buy a GPU guide 😉
-
Ellison’s $300B GPU Cluster Bet on AGI
By
–
> be me
> Larry Ellison
> own a database empire, a sailboat, and a disdain for poor people > OpenAI wants to build AGI
> needs compute
> *a lot* of compute
> like “$300B GPU cluster in a volcano” levels of compute > "Hello. I own Oracle Cloud. Also, I’m rich."
> sign $300B GPU -
40K Followers Giveaway: Four RTX 3090 GPUs
By
–
if i get to 40k followers between today and October 1st i'll give away 4 RTX 3090 GPUs to 4 random followers p.s. if you read my pinned tweet, you know how to get this done 😉
-
Incredible 800% Growth Achievement in Under Six Months
By
–
800% growth in less than 6 months, cannot believe it thank you all for choosing to hangout with me
-
Enjoying Learning and Practical Use of Technology
By
–
The nerd in me enjoyed studying it The poaster in me is enjoying using it
-
Software Engineers Fear Hardware in Cloud-First LLM Era
By
–
the amount of people saying "just use an API bro" when it comes to LLMs is insane i get it, not everyone's into gpus or hardware; but do these people realize that we're in an era where Software Engineers are legit SCARED of hardware? everything's "in the cloud" now, vendors are
-
Self-bootstrapping AI agents: Building tools with Claude
By
–
> be Anthropic
> building tools for Claude
> want it to act like a 10x engineer with a nervous system made of JSON
> realize agents are only as good as the tools you give them
> decide to write tools *with* the agent itself
> self-bootstrapping code god loop activated > first -
GPU Hardware Strategy: RTX Pro 6000 Max-Q Acquisition Guide
By
–
Incremental acquisition Easier to sell when needed At this moment if I am getting any more Nvidia GPUs I'd do the RTX Pro 6000 Max-Q, minimum, but that's because my long term plans are different from when I started And I still wouldn't recommend people to start with them