I can see that and I don't know your setup, but there are people pushing >70t/s with MTP on a single 3090, with big contexts still >50t/s. 64k is not a lot for agentic coding, and IIRC 128k with turboquant and the likes should be possible for you. r/LocalLLM/ and r/LocalLLaMA/ are worth a visit IMO.
I managed to execute with vllm successfully, but it breaks opencode on simple "what's this repo?" task. On oh-my-pi it wont event execute because omp sends multiple system prompts. I'll try with llama.cpp later and see if it works more reliably.
"[...] In the coming days, we will also open-source smaller-scale variants, reaffirming our commitment to accessibility and community-driven innovation. [...]"
In a practical sense, I'm primarily interested in small to medium sized models being open. I think that might be common sentiment.
However, my hope is that there will be at least somewhat competitive big and open models as well, from an ethical/ideological perspective. These things were trained on data that was provided by people without their consent, so they should at least be be publicly accessible or even public domain.
Qwen3.5-Plus is the largest variant of the open weight Qwen3.5 model, expanded with a 1M context window and fine-tuned on the Qwen-native harness’ specific tools.
It‘s mainly due to system requirements that Flux.2-dev doesn’t get same usage as Z-Image. A 5090 needs about a minute to generate an image with a basic workflow with Flux.2-dev. But prompt adherence and scene/character consistency in edit mode is (way) ahead of Qwen-Edit-2509 if you ask me.
I feel tired reading about him and it doesn’t even phase me anymore. It‘s just another thing I add on to the pile. Maybe it’s part of the plan to go numb to everything he does.
I still don’t understand what a “leftist” is, exactly. In my politics, I guess I’m conventionally more of a centrist. I’ve never registered as a Democrat. I tend to stay out of “woke” activism. But I’m appalled that a convicted felon and rapist who very visibly tried to steal the last election is now performing a hostile takeover of the federal government together with the richest businessman in the world. Does that make me a “leftist”? Or just, like, a totally fucking normal person?
>I’m appalled that a convicted felon and rapist who very visibly tried to steal the last election is now performing a hostile takeover of the federal government together with the richest businessman in the world. Does that make me a “leftist”?
sadly, yes. Look at the comment upstream trying to equivalate "musk fanboys" and labeling opponents as "people on the internet consumed by hate". That's the US polarization at work.
lot of apolitical people just want to sweep everything under a rug and ignore it. I realized at the beginning of the month that this isn't something to ignore, though.
Its comedic to hear that the left is the party of hate. Yes, up is down, down is up. War is peace. All that. Sure man, sure.
I somewhat agree that there's a totally out of touch disconnected sort that is bothered by being part of the body electorate, that doesn't take seriously a civic duty: would rather not pay attention, who doesn't like the conflict. I doubt your biased lopsided anecdata, doubt know many centrists changing their vote. But as the grossly unpopular & despised Project 2025 that Trump disavowed steamrolls this nation & as algorithmic AI run systems perhaps start being used by the state to mechanize programmed bias, well, those folks will be impacted deeply, and saddened, and it's unfortunate they were left slumbering & derelict from their civic duty to pay real attention, or to tune into something besides Fox News or Newsmax earlier.
I mean they make a bold statement up top just to paddle back a little bit further down with: "[…] In terms of Chinese and Cantonese recognition, the SenseVoice-Small model has advantages."
The Guardian, especially for their podcasts, is the only news website I am paying and have ever payed for. And I pay more willingly than any other newspaper would get from me for their paywall stuff. It‘s that valuable for me to support this approach.
EDIT: just found this recipe repo, may wanna give it a go: https://github.com/noonghunna/club-3090
EDIT-2: this can also shave off a lot of context need for tool calling -> https://github.com/rtk-ai/rtk
reply