I read a lot of "whengguf" type posts. I think we should sometimes stop and be grateful.
I want to say big thanks to all of the people and companies who gave us so much fun and productivity, sacrificing a lot sometimes; but also to companies who gave us models as by-product of their strategy, too.
I can miss a lot (I want to update this list if you point me at what I miss), but If I would build hall of fame, then I will put these people, companies and models (I forgot so much) there:
Meta for all LLamas up to LLama 3.3
Mistral for Mixtral 8x7B, Mistral Large, and Mistral Medium 3.5
OpenAI for Whisper models, GPT-OSS-20B/120B and distill foundation of Chinese open-weight models
Google for Gemma models, which focus on different things than mainstream (e.g. medical images and a lot more)
DeepSeek for DeepSeek-V2/V3/R1 and V4
Alibaba for Qwen models, especially Qwen2.5-32B Coder, QwQ, Qwen3.x
Georgi Gerganov and the whole llama.cpp team, together with ikrakow and rest who departed
TheBloke, bartowski, unsloth, mradermacher and countless people who made and make quants
HuggingFace for hosting all this petabytes of models happiness
"Attention is all you need" paper authors
RAG concept authors
Honorauble mentions:
MoonshotAI for Kimi 2.x models
Minimax models (for good coding alternative)
LMStudio (for those who can't llama server)
Open WebUI (for trying to make OSS LLM administration).
[link] [comments]