Virtus

165 posts

Virtus banner
Virtus

Virtus

@Zenitshi

Smile More

Katılım Kasım 2024
25 Takip Edilen17 Takipçiler
Sabitlenmiş Tweet
Virtus
Virtus@Zenitshi·
Axo: v2.0 is out 🎉 Its a desktop app designed to streamline your workflow by providing powerful voice dictation, intelligent text refinement, and AI prompt engineering capabilities. It listens to your voice, transcribes it. Open source & Free & Local github.com/Zenitshi/Axo
GIF
Virtus tweet media
English
0
0
2
301
Virtus
Virtus@Zenitshi·
@alabhya2007 @Pirat_Nation with copying u learn after that u innovate its like in arab and persian golden age europe copied their ways and then added their sauce and other nations on the past its okay to copy as long as u innovate afterwards
English
1
0
1
160
ARIRANG ⁷
ARIRANG ⁷@alabhya2007·
@Pirat_Nation China was, is and will be copy cats. They copy other companies technologies and say we built it. Such a foolish country.
English
2
0
4
1.7K
Pirat_Nation 🔴
Pirat_Nation 🔴@Pirat_Nation·
A former Samsung researcher has been sentenced to 7 years in prison in South Korea, he was convicted of leaking key DRAM semiconductor technology to a Chinese company. The 56-year-old man used to work at Samsung Electronics, after he left the company, he joined ChangXin Memory Technologies, or CXMT, in China. This case involves 10 people who were charged, officials believe the leaked details helped China move faster on its own memory chips, including those used for artificial intelligence. Reports say the researcher and another ex-Samsung employee received millions of dollars for the information. South Korea treats its semiconductor know-how as vital to the nation’s economy.
Pirat_Nation 🔴 tweet mediaPirat_Nation 🔴 tweet media
English
39
83
1.3K
53.3K
Jamie Day
Jamie Day@JamieDay719679·
@darrenangle Has anyone actually run this? It says is occupies like 8 gb of vram when I do.
English
1
0
2
189
darren
darren@darrenangle·
an 8B near qwen level with less than 2GB memory is frankly ridiculous
PrismML@PrismML

Today we’re announcing Ternary Bonsai: Top intelligence at 1.58 bits Using ternary weights {-1, 0, +1}, we built a family of models that are 9x smaller than their 16-bit counterparts while outperforming most models in their respective parameter classes on standard benchmarks. We’re open-sourcing the models under the Apache 2.0 license in three sizes: 8B (1.75 GB), 4B (0.86 GB), and 1.7B (0.37 GB).

English
4
12
285
34.1K
Virtus
Virtus@Zenitshi·
@LuCaPloo @scaling01 As if Opus 4.6 is far ahead compared to Opus 4.5, I don't find a lot of difference if its at the same level of opus 4.5 with that price why would u refuse it, opus 4.5 level is already enough for any vibe coder or dev to build their software.
English
0
0
0
135
Sound
Sound@LuCaPloo·
@scaling01 haha lol they are comparing it to the SO VERY OLD, insanely outdated "opus 4.5". I can't believe the data about this model is still even available! It's been so long!
English
2
0
0
1.1K
Lisan al Gaib
Lisan al Gaib@scaling01·
GLM-5 Benchmarks
Lisan al Gaib tweet media
English
25
39
524
110.5K
Virtus
Virtus@Zenitshi·
@ollama @bfl_ml i thought there's fp4 in z image turbo that's too bad but good update, i definitely didn't expect this one from ollama nice
English
0
0
0
209
ollama
ollama@ollama·
Ollama is here with image generation! ollama run x/z-image-turbo ollama run x/flux2-klein In the latest release we've added experimental support for @Ali_TongyiLab Z-image-turbo @bfl_ml Flux.2 Klein! (macOS with Windows and Linux coming soon) See examples 👇👇👇
ollama tweet mediaollama tweet mediaollama tweet mediaollama tweet media
English
51
230
1.3K
92.8K
Landon Garrison
Landon Garrison@lando_garrison·
GLM 4.7 and MiniMax 2.1 are the first time I genuinely feel Chinese made models have surpassed American models. Both models may not be better in every way, but they feel infinitely easier to use and get what you want, plus cheaper (this is from my experience)
English
38
30
746
49.6K
Virtus
Virtus@Zenitshi·
@DilyanDimitrov @jonathan_wilke @nextjs depends on the version ur using and also u mean dev or prod read 16.1 or 16 next doc for faster navigation for extra speed use bun instead of nodejs
English
0
0
1
383
Virtus
Virtus@Zenitshi·
@braelyn_ai except they didn't really built it they bought a fork of windsurf and added their feature like agent manager i hope they make it more stable and then adding new features such as custom modes, context engineering, better tab completion and what not
English
2
0
0
116
Braelyn ⛓️
Braelyn ⛓️@braelyn_ai·
can someone build Antigravity but without Google?
English
123
6
324
38.1K
Virtus
Virtus@Zenitshi·
@abhitwt its too soon to decide for antigravity: antigravity have potential cursor already in much more stable version where antigravity isnt its not fair to compare right now but if u want to pick a winner cursor cuz its more mature but for affordability antigravity
English
0
0
1
26
Abhishek B R
Abhishek B R@abhitwt·
pick one for the rest of ur life
Abhishek B R tweet mediaAbhishek B R tweet media
English
337
9
630
87.1K
Virtus
Virtus@Zenitshi·
@AGI_is_woke @scaling01 they showed agent workflow of gemini 3 and its kinda shows something but i got u
English
0
0
0
79
AGI ~ May 2027
AGI ~ May 2027@AGI_is_woke·
@scaling01 there's a reason they didn't show gemini 3 pro in the computer use benchmark
English
1
0
2
956
Lisan al Gaib
Lisan al Gaib@scaling01·
ByteDance released Seed 1.8 they have really good browser/computer use
Lisan al Gaib tweet mediaLisan al Gaib tweet mediaLisan al Gaib tweet media
English
7
45
477
38.4K
Virtus
Virtus@Zenitshi·
on ai studio
Virtus tweet media
Italiano
0
0
0
23
Virtus
Virtus@Zenitshi·
New model of gemini ?!!!!! Gemini Robotics-ER 1.5 is a (VLM) that brings Gemini's agentic capabilities to robotics. It's designed for advanced reasoning in the physical world, allowing robots to interpret complex visual data, perform spatial reasoning. ai.google.dev/gemini-api/doc…
English
1
0
1
70
Ivan Mehta
Ivan Mehta@IndianIdle·
Wispr Flow now has localized Indian Indian pricing of ₹320 ($3.5) per month instead of $12 per month for annual billing.
Ivan Mehta tweet media
English
42
18
591
114.4K
Virtus
Virtus@Zenitshi·
@ChoogchoogX @theo its also a windsurf fork the problem is that they should know how to deal with each model correctly gemini is so talkative well its free so why we need to complain
English
0
0
0
58
🌈חוגי
🌈חוגי@ChoogchoogX·
@theo It's a VSCode fork, it's "Gemini 3 pro" which sometime feels stupider than 2.5 flash that's the problem. but the amount of tokens they give out is insane. more than GPT's 200$ for 20$
English
1
0
5
1K
Virtus
Virtus@Zenitshi·
GPT 5.1 VS GPT 5.2 that's cool
Virtus tweet media
English
0
0
0
46
Virtus
Virtus@Zenitshi·
@victoria_framer the whole point of websites is info if u present it like shopify or comet or luma its gonna be like a video and i think that's what it will be if they focus more on utility it will be like a video displaying ur products and info needed for the user in interactive way
English
0
0
0
92
Virtus
Virtus@Zenitshi·
@eliebakouch if that's the case can u run it on one rtx 5090 ?
English
0
0
1
324
elie
elie@eliebakouch·
nice that it's open weight, but comparing dense vs moe models and only looking at total params is pretty unfair, if you look at active params instead of total params it's a different story: - GLM 4.6 (32B): 74% fewer - Minimax M2 (10B): 92% fewer - K2 thinking (32B): 74% fewer - V3.2 (37B): 70% fewer size (both total or active!) is not the right metric here, we should have the same graph with speed on vllm / sglang
elie tweet media
Mistral AI@MistralAI

Introducing the Devstral 2 coding model family. Two sizes, both open source. Also, meet Mistral Vibe, a native CLI, enabling end-to-end automation. 🧵

English
18
18
291
54.8K
Virtus
Virtus@Zenitshi·
@heyruchir @suchenzang sonnet param size is not public heck the whole anthropic models param size is not public its all speculation and guesses
English
0
0
0
871
Virtus
Virtus@Zenitshi·
It would've been nice if someone made a benchmark for all the local models, the small models that can handle laptops or PCs, and make a comparison based on the VRAM and the performance.
English
0
0
0
23

Keşfet

@alabhya2007 @Pirat_Nation @JamieDay719679 @darrenangle @LuCaPloo @scaling01 @ollama @bfl_ml