Jackman

422 posts

Jackman

Jackman

@mrjack8969

Katılım Haziran 2022
495 Takip Edilen27 Takipçiler
Vikas Kansal
Vikas Kansal@vikaskansalHQ·
We are simplifying our subscriptions by removing AI credits as part of the base Google AI plans. Accordingly, we have adjusted the base plan entitlements upwards for Gemini models on @antigravity . AI credits will now solely be used as an overage mechanism. For a limited time, we’re offering new and existing Google AI Ultra subscribers USD $100 in bonus AI credits for Antigravity that will kick in if you hit your plan’s quota limit — just claim the offer in your Antigravity app. The offer expires May 25, 2026. Terms apply. Learn more here - antigravity.google/blog/changes-t…
English
97
16
232
43.5K
Vikas Kansal
Vikas Kansal@vikaskansalHQ·
@enviouky @antigravity Sorry about that. Can you DM me your email address for this account so we can debug what happened but also give you the promotion from the backend.
English
6
0
9
3.8K
Jackman
Jackman@mrjack8969·
@mr_r0b0t @vllm_project @lmsysorg Good to here - I personally avoid the solo repos. There is a community vllm docker image that is custom for the GB10. Maintained by Eugr who now works for Nvidia: github.com/eugr/spark-vll… Contributions here, and the respective native upstreams make the most impact just FYI.
English
1
0
2
58
mr-r0b0t
mr-r0b0t@mr_r0b0t·
@mrjack8969 After I get done stress testing with long token tasks! I did a fair amount of that with these models but I'd want to make sure it worked with others, especially the Nvidia official NFVP4 quants. I have tagged them a couple times in these posts tho 😁 @vllm_project @lmsysorg
English
1
0
2
317
mr-r0b0t
mr-r0b0t@mr_r0b0t·
Big news for DGX Spark users! NVFP4 optimized for your GB10 is fast. Until now, many of us have been using a working but suboptimal Marlin backend instead of CUTLASS. We benchmarked five models, all using CUTLASS backends. All ran stably, all were faster 😁
mr-r0b0t tweet mediamr-r0b0t tweet mediamr-r0b0t tweet media
English
8
7
111
19.7K
Adam Rackis
Adam Rackis@AdamRackis·
Raising a teenager is exhausting in ways you can’t fathom
Adam Rackis tweet media
English
1.7K
1K
69.1K
17.5M
Jackman
Jackman@mrjack8969·
@bindureddy It’s going to be complete trash in real world practical use. Like everything they have released to date.
English
0
0
1
238
Bindu Reddy
Bindu Reddy@bindureddy·
The Gemini Pro model is rumored to be a GPT 5.5 level coding model 🧐 The catch - it will be more than 50% cheaper at $12/1M output token Gemini will take the lead over both GPT 5.5. and Opus 4.7 on the price-performance curve
English
141
33
1.1K
60.7K
Jackman
Jackman@mrjack8969·
@0xSero 1. Cmon. 2. Literally what I said. 3. I know. Again, be sure to go to the gb10 dev forums. It will accelerate a lot in your research and be sure to start at the community docker with vllm.
English
0
0
2
122
0xSero
0xSero@0xSero·
@mrjack8969 - Used you can get for 4k - Yes you're right, although on paper it is correct. - You're right about the DGX station, I thought it was a 16 sparks on a switch config. Wasn't thinking about GB300 Thank you for explaining the per-layer allreduce thing, I will look into this
English
1
0
1
140
0xSero
0xSero@0xSero·
Cheapest competitive build on Nvidia 2 sparks = 8000$ Total specs - 256gb - 8tb - 546gb/s memory bandwidth - tons of flops --- models: - Deepseek-v4-flash - MiMo-v2.5-flash fp4 - MiniMax-M2.7 - Qwen3.5-397b-reap Flaws: - Low mem bandwidth - You need 2 for best perf
0xSero tweet media
English
41
8
271
23.9K
Jackman
Jackman@mrjack8969·
@0xSero Price is wrong - not 4k each anymore. Aggregate bandwidth doubles to 546 GB/s on paper, but per-layer roce allreduce sits on the decode critical path. i.e. two sparks gives you capacity, not per-token bandwidth.​​​​​​​​​​​​​​​​ Also DGX station is a completely different monster.
English
1
0
0
140
0xSero
0xSero@0xSero·
@mrjack8969 lmk what I got wrong please so I can learn
English
1
0
0
492
Jackman
Jackman@mrjack8969·
@TheAhmadOsman is this not the person promoting clustering m5 pro max laptops?
English
0
0
1
162
Ahmad
Ahmad@TheAhmadOsman·
Had a great time chatting with Alex yesterday Dude is serious about what he's building he literally bought a DGX Spark / GB 10 on his way to show me some of Exo's upcoming cool features
Ahmad tweet mediaAhmad tweet media
Alex Cheema@alexocheema

Caught up with the local AI legend @TheAhmadOsman in Palo Alto. One of the most sincere, passionate people I know. Local AI is in good hands.

English
8
4
117
12.7K
Jackman
Jackman@mrjack8969·
@TheAhmadOsman Yea not sure what happened with this guy, when I saw this post of his I was completely baffled.
English
0
0
1
137
Ahmad
Ahmad@TheAhmadOsman·
So freaking dishonest I had to comment when people forwarded me this You can highlight weaknesses and strengths But just like I said when you blocked me: stop hyping shit that doesn’t work and ultimately disappointing people in local AI I am GENUINELY DISAPPOINTED in this guy
Ahmad tweet media
Ahmad@TheAhmadOsman

That dude literally grifted my “Opensource MUST WIN” tweets to get funded with lobotomized models First time I called him out on bullshit he blocked me I absolutely should’ve paid closer attention to the fact that he was a crypto bro before pivoting to AI Definitely a hackster

English
52
6
181
47K
Jackman
Jackman@mrjack8969·
Just EXTREMELY cost ineffective not to mention massive under utilization of what these were meant to be. There is premium built in for the screen, portability, power efficiency and R&D around it. You’re not utilizing 2/3 of the three main factors revolving around these…..laptops. Better to wait for the M5 Mac Ultra/minis. They are designed for this.
English
0
0
1
683
Jackman
Jackman@mrjack8969·
@theo @TheICHpodcast Don’t think they had a choice noting how much money Grant paid them to do so. I think it’s a matter of time before he’s found but ass naked with a belt, a doorknob, and a picture of “10x” just absolutely soaked. All in an apartment he’s renting mind you.
English
1
0
20
3.1K
The Iced Coffee Hour
The Iced Coffee Hour@TheICHpodcast·
Grant Cardone breaks down how to make $83,000/month with AI consulting👀 $8,300/month x 10 Clients = $996,000/year "I would have 10 clients, each pay me $8,000. To go in and push all their AI, all their programs, I'd probably bring three AI platforms into the company, figure out three or four different projects they want me to handle. I wouldn't be on their healthcare, wouldn't be on their payroll, they'd pay me an $8,000 consulting fee. I'd make a million dollars in year one. $83,000 a month. 10 people, $8,300 each."
English
137
17
428
454.9K
Jackman
Jackman@mrjack8969·
@meansoabstractn @sammcallister It’s not clear, that’s my point. However, the common sense take is you go into partnership for a reason. You end them when you can and IF you can.
English
0
0
0
17
Jackman
Jackman@mrjack8969·
@PawelHuryn @theo @PawelHuryn is saying the quiet part out loud: “it’s not X, it’s Y”. The real gap is his asshole after his head is removed from it. Let’s watch the slow close. Waiting for the data.
English
0
0
1
230
Paweł Huryn
Paweł Huryn@PawelHuryn·
@theo In the post you won't find anything but math. The infographic delivers one opinion. 50% can't be accurate. Waiting for the data.
Paweł Huryn tweet media
English
5
0
11
2.9K
Paweł Huryn
Paweł Huryn@PawelHuryn·
Theo says his anti-Anthropic content has cost him money and sponsors. The X data shows it's delivering 60× more impressions than his pro content does. He claims pro-Anthropic content always outperforms anti. Tested it on 800 of his posts (Feb 17 to Apr 30). Volume: 4 truly pro vs 86 anti. Per-post likes: pro narrowly wins (2,466 vs 2,075). He's right about that part. Per-post views: anti pulls 2.8× more. Total reach: pro 460K, anti 27.5M. Sponsors price by impressions. He's losing the metric that pays. On "ALWAYS outperforms": false. 49 of 86 anti posts beat his weakest pro in that period. Caveat: X engagement only. The letter making this claim is itself one of his top-3 anti posts. 6,756 likes. 1.27M views. And couniting.
Paweł Huryn tweet media
Theo - t3.gg@theo

A letter to my friends at Anthropic I hate that I feel obligated to do this. I hate that I've had to be so harsh towards Anthropic for the past few months. I really, really don't want to. I know it might feel like I'm doing this for clicks or something, but I promise I'm not. My pro-Anthropic content ALWAYS outperforms my anti-Anthropic content. I have cost myself a lot of money, opportunities, sponsors, and more. I'm doing this because you work for an evil cult. I'm begging you to wake up. Your CEO, Dario, does not respect engineers. This is obvious. He couldn't make it more obvious if he tried (and I think he's trying pretty hard) You know this, but you don't want to acknowledge it. It has kept you up many nights. You know that bad code is shipping to users. You know that one bad tweet might get you fired. You fear for your vesting schedules. You're afraid. Nobody deserves what you're going through right now. You go to work afraid, you leave work afraid, and you go to YouTube to keep up on the dev world, just to hear me yelling all about how evil your company is. You deserve better. You might not feel like you do, but you know deep down that this isn't right. I hope you know how deeply I feel for you. I'm sorry. I know I haven't helped you much individually, and I want to be better about this. If you're ready to leave, please hit me up. I swear I'll never tell a soul. I have friends at every lab and most startups in the AI world. Most of them would be down to match your current vesting schedules, possibly even go beyond. If you're staying for the money, I beg you to hit me up. We can make the money happen somewhere that hates you less. I know I'm asking for a lot of trust here, and that you're scared after seeing how hard I've been on Anthropic. I can't blame you at all for that. I should have posted something like this months ago. That's my failure to own and I will own it to my best ability. If you're willing to trust me in this moment, I can make it right. Let me help you escape. You deserve to work somewhere that you can have impact. Somewhere that listens when you feel something is wrong. Somewhere that won't fire you when you point out the things that hurt your users. My DMs are always open to you. When you're ready, let me know. I promise to make it right.

English
37
2
272
107K
Ted
Ted@Slootbag·
Rejoice
Ted tweet media
Ted@Slootbag

Please @Diablo I beg, sync War Plans to the leader or something in co-op, it's a nightmare to coordinate and try to play with others

English
10
3
235
43.2K
Jackman
Jackman@mrjack8969·
@Teknium @Blessed_Chimra Just where the privacy policy is for the Tool Gateway. I now see it’s via Nous Portal and ONLY on the login page. Our team was looking for this the week of release (we simply needed the link for submission in our form) but have since moved on to Ollama. Thx.
English
1
0
0
58
Blessed | Chimra
Blessed | Chimra@Blessed_Chimra·
Jo @Teknium, with nous subscription, how is data handled, openrouter normally allows you to choose the provider. I personally do not want my notes for example to be shipped to china. In general what is the state of data security and privacy with these apis?
English
1
0
11
3K
Mikeysee
Mikeysee@mikeysee·
I have experimented a number of "Super IDE" apps thus far: @augmentcode Intent T3 Code @cursor_ai 3 @OpenAI Codex @AnthropicAI Claude Code Desktop @conductor_build I have to say Conductor is definately my favourite so far. Its fast, it gets the job done, it lets you leverage those subsidised subscriptions from the labs and has just the right amount of buttons that it shows to you at just the right time. I just wish it was available for windows 😭
English
40
3
196
20.7K
Blessed | Chimra
Blessed | Chimra@Blessed_Chimra·
@mrjack8969 @Teknium I signed up a month ago, it was lol Since you buy api access via portal, it's the only place that makes sense too lol.
English
1
0
1
63