I can run Gemma4, Qwen 3.5, Dolphin3 and some others all 9B or 12-14B, but with patience and without GPU using ollama. I want to try llama.cpp, looks better.
I have a RX 580 8GB of VRAM, but it's better don't use it 🤣
I use to create some tiny code, corrections and some scripts for automation, shortcuts. I like Qwen for it.
Answers for questions and translations, Gemma wins.
Dolphin and Llama are faster to response, nice to chatbots. I think.
baunilha / [Baunilha]
npub1cp…2nqrh
2026-10-23 20:54:25 UTC
in reply to nevent1q…dey6
Author Public Key
npub1cp7q2jvfe9ehyp9zyn7szzfdslpu4jex0y3uxx9qh6zdde6ush6s02nqrhSeen on
wss://relay.damus.ioPublished at
2026-10-23 20:54:25 UTCEvent JSON
{
"id": "000000bf8421c129f888ee9017e90e67ec382ae74d9829f0413234bbf98400b3",
"pubkey": "c07c054989c9737204a224fd01092d87c3cacb267923c318a0be84d6e75c85f5",
"created_at": 1792788865,
"kind": 1,
"tags": [
[
"e",
"8373c57c8b13e849fb534a3caa377473a45be5db1a1d8c359efc342ef6e2ceae",
"wss://nostr.wine/",
"root",
"4d7842051782e0d3feb034d150adc2b6bae4ee3b49786793bffa468b6f5b96b3"
],
[
"p",
"4d7842051782e0d3feb034d150adc2b6bae4ee3b49786793bffa468b6f5b96b3",
"wss://relay.damus.io/"
],
[
"client",
"Amethyst"
],
[
"nonce",
"300]Xm",
"24"
]
],
"content": "I can run Gemma4, Qwen 3.5, Dolphin3 and some others all 9B or 12-14B, but with patience and without GPU using ollama. I want to try llama.cpp, looks better.\n\nI have a RX 580 8GB of VRAM, but it's better don't use it 🤣\n\nI use to create some tiny code, corrections and some scripts for automation, shortcuts. I like Qwen for it.\nAnswers for questions and translations, Gemma wins.\nDolphin and Llama are faster to response, nice to chatbots. I think.",
"sig": "154ef91b835c44908fecdb7ceeb4f83c47e2f2dfaacccca028cf0a33fb73a839a1b5a41a6181cea66df4ce0a423d9c6c4582f10619a1fc117be35bc9da7f3f7e"
}