temp

882 posts

temp

temp

@prof_intern

I like data

Sumali Kasım 2025
52 Sinusundan180 Mga Tagasunod
temp
temp@prof_intern·
Can't wait to see a dedicated coding model from grok that can rival GPT5.4 or Opus while being slightly more cost efficient. Cost efficiency is the way to go right now, with Opus 4.7 being even more token hungry than Opus 4.6. Gemini could do literally anything. They seem like they're not even trying
Mark Kretschmann@mark_k

Grok 4.3 (beta) has been released by @xai 🔥🔥 The new version of Grok is currently available to the highest tier plan SuperGrok Heavy. With 1T params Grok 4.3 is twice as big as Grok 4.20, and was also trained much longer. It should provide a decent jump in performance.

English
0
0
3
100
temp
temp@prof_intern·
@Azzapp_LoL Except where the system doesn't work and doesn't detect soft inting? Or you just play against enemies better than you and want to simply lose faster
English
0
0
1
119
Azer Dugalić
Azer Dugalić@Azzapp_LoL·
Surrender basically has no purpose in ranked anymore. Don't connect for a minute = remake Go afk or dc = game terminated Somebody inting = game terminated All of the reasons why surrender had to exist back in the day are gone. System is much more robust now and can figure out deeply unfair scenarios, everything else is playable.
Azer Dugalić tweet media
English
104
47
2.1K
363.6K
temp
temp@prof_intern·
Can't agine the limit usage on Opus 4.7 looking at how token hungry is Opus 4.6. It will be unusable
English
0
0
3
17
temp
temp@prof_intern·
@ravikiran_dev7 36 is a rookie number, just an hour of applying
English
0
0
1
24
Ray🫧
Ray🫧@ravikiran_dev7·
A senior engineer created a fake junior dev to test hiring advice. The profile had: - good portfolio - active GitHub - senior-level projects - even faked YOE He ran the experiment in the Netherlands, a country known for its developer shortage. Applied to 36 jobs and didn’t get a single interview. getting interviews feels harder than the actual job rn !!
Ray🫧 tweet media
English
50
15
230
49.6K
temp
temp@prof_intern·
temp tweet media
Bridgebench@bridgebench

Gemini 3.1 Pro ranks dead last among frontier models on BridgeBench Reasoning. Behind Grok 4.20, GPT 5.4, Claude Opus 4.6, Qwen 3.6 Plus, MiniMax M2.7, Claude Sonnet 4.6, and GLM 5.1. Google's flagship model can't even beat a free Chinese model on grounded reasoning. This is why I cancelled my $250/month Google AI Ultra subscription. Gemini CLI was unreliable. The model is mid. The infrastructure is worse. Google has the compute. Google has the data. They just can't ship a competitive coding model.

ZXX
0
0
2
26
Mgoes (bio/acc 🤖💉)
Mgoes (bio/acc 🤖💉)@m_goes_distance·
All you have to do is survive the next 5 years this is what the world looks like in 15-20 years: -humans hitting 130 routinely, 150 if you don't get bored early -jet suits replacing the morning commute -organs printed to your exact spec on demand -centenarians who are sharp, dangerous, and still running things -body and brain upgrades you walk into a clinic and get -cognitive enhancements that make your 25-yo self look slow -avatars that work, travel, and take meetings etc the tech exists, just not evenly distributed yet please don't quit in the next 3-5 years. bio/acc.
English
138
102
1.2K
172.9K
Shubhank K.
Shubhank K.@shubhanalytics·
Let’s connect, If you are into: Data Science Data Analytics Data Visualisation Business Intelligence Business Analytics NLP LLM RAG AI/ML #LetsConnect #ConnectandEngage
English
26
1
27
909
temp
temp@prof_intern·
???
temp tweet media
ZXX
1
0
3
15
temp
temp@prof_intern·
@RiotPhroxzon @drewlevin The matchmaking is great, but I feel like there might be something wrong here. Could anyone tell me if my feeling is correct?
temp tweet media
English
0
0
3
7
temp
temp@prof_intern·
@khushiirl Google is and was always behind
English
0
0
1
17
khushi.vy
khushi.vy@khushiirl·
Still I’m wondering why Gemini fails against Claude and GPT ?? - Owns Chrome - Backed by Android - Holds ~95% search history - Stores most search results - Google has the biggest user data - Even incognito data isn’t fully private So what’s the problem?
English
83
3
240
23.2K
temp
temp@prof_intern·
> be a big size company > look for ways to save money > AI > layoffs > people with no jobs have no money > people with no money don't buy your products > you sell less > you earn less > you collapse and go bankrupt > profit
temp tweet media
English
0
0
3
11
Prompter
Prompter@PromptLLM·
Gemini 3 really had the biggest fall from grace ever. It is unusable
English
117
17
1K
128.4K
temp
temp@prof_intern·
Mythos will not finish a plan without hitting the limits at 200$ tier. Meanwhile, GPT 5.4 lets you actually work, achieving comparable results
temp tweet media
English
0
0
2
6
Harshith
Harshith@HarshithLucky3·
First Google reduced Antigravity rate limits for Pro users Next Anthropic did the same for Claude web/app and Claude code Now, OpenAI is doing the same with Codex. After the release of the $100/month plan, OpenAI reduced the rate limits for Plus users and it became worse. It's really getting ~1/3 of previous usage
English
62
56
735
76.9K
Prasenjit
Prasenjit@Star_Knight12·
guess the AI model
Prasenjit tweet media
English
296
7
831
212.6K
temp
temp@prof_intern·
@haider1 At this point people who don't need 1000$ sub may just use 20$ one with bit worse models, small group of people need opus 4.6 . Sonnet does great job in most tasks, people just think they need it
English
0
0
1
387
Haider.
Haider.@haider1·
PREDICTION it's very likely that within a year, openai and anthropic will have subscriptions that cost $1,000 per month or at least $500 per month anthropic will probably do this first. $20 per month plan will become much more similar to the free plan
English
131
15
470
23.3K

Tuklasin

@Azzapp_LoL @ravikiran_dev7 @m_goes_distance @shubhanalytics @labul_gong @RiotPhroxzon @drewlevin @khushiirl