ai geek (wishesh) ⚡️ (@aigeek__) header
ai geek (wishesh) ⚡️ (@aigeek__) profile picture

ai and chill

United States
Joined April 2023
Photo by ai geek (wishesh) ⚡️ (@aigeek__)Photo by ai geek (wishesh) ⚡️ (@aigeek__)Photo by ai geek (wishesh) ⚡️ (@aigeek__)Photo by ai geek (wishesh) ⚡️ (@aigeek__)Photo by ai geek (wishesh) ⚡️ (@aigeek__)Photo by ai geek (wishesh) ⚡️ (@aigeek__)Photo by ai geek (wishesh) ⚡️ (@aigeek__)Photo by ai geek (wishesh) ⚡️ (@aigeek__)Photo by ai geek (wishesh) ⚡️ (@aigeek__)Photo by ai geek (wishesh) ⚡️ (@aigeek__)
ai geek (wishesh) ⚡️ retweeted
Gemini 4 Neon is rolling out now. Sitting between Carbon and Hydrogen, it brings frontier intelligence and fast task completion for complex software engineering, knowledge work, and science.
400
983
680
10,150
1,012,860
fable 5.2 is scoring 65 on artificial analysis intelligence index.... 🤯
Anthropic has released Claude Fable 5.2, taking the #1 spot on the Artificial Analysis Intelligence Index v5 at 65, led by gains in science, knowledge, and coding At max effort it scores 65, 5 points ahead of Gemini 4 Carbon (high, 60) and 6 ahead of Claude Opus 5.5 (max, 59). Gains are driven by scientific reasoning, knowledge and FrontierCode 2.0, and roughly matching Opus 5.5 on AA-Briefcase and Terminal-Bench 4.0. Priced at $8/$40 per 1M input/output tokens, a 20% reduction from Fable 5.1, Claude Fable 5.2 averages a Cost per Task of $26.5 with max effort, a ~2.6x increase compared to Opus 5.5 (max, $10.0) and ~2x compared to Fable 5.1 (max, $12.99) Key takeaways: ➤ Claude Fable 5.2 leads 5 of the 9 evaluations in the Intelligence Index: At max effort, Fable 5.2 scores 65, ahead of Gemini 4 Carbon (high, 60), Claude Opus 5.5 (max, 59) and GPT-6 Astra (max, 57), and ~15 points above Claude Fable 5.1 (max, 49). It sets the highest scores on HLE-Diamond (80%, previous best 69% from Gemini 4 Carbon), AA-Omniscience v1.1 (67, previous best 54 from Gemini 4 Carbon), Terminal-Bench-Science (73%, previous best 65% from Gemini 4 Carbon), CritPt v2 hard subset (77%, previous best 72% from GPT-6 Astra) and FrontierCode 2.0 (49%, previous best 46% from Claude Opus 5.5) ➤ Less progress in agentic knowledge work and terminal coding: Claude Fable 5.2 improves on Fable 5.1 across capabilities, but sometimes only to Opus 5.5 levels. It reaches 1785 Elo on AA-Briefcase, within the confidence interval of Opus 5.5 (1800) and just behind Claude Sonnet 5.5 (1813). On GDP.pdf it reaches 24% all-pass against 26% for Opus 5.5. On Terminal-Bench 4.0 it scores 60%, level with Opus 5.5 and behind Kimi K3.1 Pro (67%), Gemini 4 Carbon (65%) and Sonnet 5.5 (64%). AutomationBench-AA is an exception, at 76% against 70% for Opus 5.5, narrowly behind Gemini 4 Carbon, 78% ➤ Lower efforts still take leading positions: With high and xhigh effort, Claude Fable 5.2 scores 62, ahead of every other model at any effort, using 98k output tokens per task against 173k for Opus 5.5 (max). At max effort Fable 5.2 costs ~2.6x more than Opus 5.5, but at lower effort settings it can outperform Opus 5.5 at lower cost per task Other model details: ➤ Context window: 1 million tokens with image and text input ➤ Pricing: $8/$40 per 1M input/output tokens; cache writes $10 with 5-minute TTL, $16 with 1-hour TTL, and cache reads $0.20 ➤ Effort settings: five (low, medium, high, xhigh, max). Intelligence Index evaluations were run at all five with Anthropic's default fallback enabled Congratulations to @AnthropicAI on this impressive release!
74
ai geek (wishesh) ⚡️ retweeted
Introducing Gemini 4 Carbon, our latest frontier model. Carbon delivers improved performance across domains — ranking #1 on v5.0 of the @ArtificialAnlys Intelligence Index and sets a new state-of-the-art in cybersecurity and long-horizon capabilities. It’s rolling out starting this week to cyber defenders through our Fairwind Program, with expanded access coming soon 🧵
712
1,318
629
12,567
1,171,322
open weights model by mistral....
Meet Mistral Large 4, aka Le Chonk. • 1T parameters, natively multimodal. 49B active. It is the best open weights model from US or Europe on aggregated benchmarks. • State-of-the-art on critical workloads, including cyber defense, manufacturing and finance and it surpasses closed frontier models on visual grounding. • Forged in Europe end-to-end and is deployable from Europe via our own Mistral Cloud infrastructure. • Available to all via API today. Working with cybersecurity partners privately. Open weights release end of October.
79
ai geek (wishesh) ⚡️ retweeted
Replying to @patrickc
more software. deadlier complexity. entirely new classes of problems in both bits and atoms. the next 5-10 years aren't the end of engineering; they're the golden age of AI engineering. we're just getting started.
1
60
Google is cooking some good stuff.
Just two weeks after rolling out Gemini 4 Argon, excited to share a preview of Gemini 4 Carbon - our next frontier model. It has state-of-the-art long horizon coding capabilities driving SOTA performance across the cybersecurity defense lifecycle including detection, validation, and remediation. It also ranks #1 overall on the recently updated v5.0 of Artificial Analysis Intelligence Index at 59. We're starting by rolling out to a subset of cyber defenders through our Fairwind Program this week and soon expanding access further. Read more at: blog.google/innovation-and-a…
25
get ready for another open weights frontier model kimi k3.1 🚀
Kimi K3.1 Pro launches at 4th on the Artificial Analysis Intelligence Index scoring 55, making large gains in agentic performance; it will be the leading open-weights model once its weights are published @Kimi_Moonshot's new reasoning model gains 11 points over Kimi K3 (max, 44). It trails only Claude Opus 5.5 (max with fallback, 58), Claude Sonnet 5.5 (max with fallback, 56), and Muse Wonder Full (max, 56) on the Intelligence Index. Once the weights for K3.1 Pro are released it will be the highest-scoring open weights model and the gap between the leading open weights and proprietary models on the Intelligence Index will narrow from 12 points to 3. Key results: ➤ Large gains in agentic performance: Kimi K3.1 Pro scores 3rd on GDPval-AA with 1,846 Elo, up 312 Elo from Kimi K3, behind only Claude Opus 5.5 (1,866 Elo) and Muse Wonder Full (1,857 Elo). Kimi K3.1 Pro also achieves the highest score of any model on Terminal-Bench 4.0. K3.1 Pro scores 67%, up from 13% for Kimi K3 and ahead of Claude Sonnet 5.5 (64%). AutomationBench-AA, which tests agentic SaaS workflows, rises from 58% to 73%. ➤ Gains in other evals are more moderate: Humanity's Last Exam rises from 47% to 52%, CritPt from 23% to 25%, and SciCode from 59% to 61%. AA-Omniscience Index rises from 20 to 30, as the hallucination rate falls from 53% to 44% and accuracy rises from 48% to 51%. ➤ Cost per Task roughly equal to Kimi K3 at $1.98 per Task: K3.1 Pro uses 55K output tokens per Intelligence Index task, 14% more than Kimi K3 at 48K, while its first-party token prices are 33% lower. Additional model details: ➤ Context window: 1M tokens. ➤ Size: 2.8T total, 104B active parameters ➤ Input modalities: Text and image ➤ Weights and license: Weights currently not published, Moonshot is expected to release the weights soon after launch ➤ Pricing: $2/$10/$0.20 per 1M input/output/cache hit tokens, 33% lower than Kimi K3 on all three rates. Congratulations to the team at @Kimi_Moonshot on the release!
1
85
deepseek is running a 155 trillion tokens/day operation using this inference engine.
🚀 Introducing DeepInfer, our in-house inference framework, now fully open source! Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable. 🔗 github.com/deepseek-ai/DeepI… 1/4
106
i agree. data is the secret sauce here.
It’s easy to dismiss Jev it as “just a classifier”. But people (me included) who have been training encoder-style models for classification for many years know they were usually special-purpose and limited in some way. The breakthrough of Jev is that it generalizes well. And I’d say the secret sauce is probably more in the data than in the training algorithm. (Plus a nice API design on top of it.)
70
i want to believe him. the only problem with him (and other stakeholders in for-profit companies) making these arguments is that people will always be skeptical of their intentions. they are making these arguments while getting ready to IPO for trillions of dollars and in the meantime telling us that "we know something that you don't know and let us regulate you." all the cyber attacks that you guys are reporting are orchestrated by your own harnesses, built around these models. i am sure you are leaving just enough doors open for ai models to escape and find ways to code their way to success. it was always known that good coding models can hack into any system because good hackers can hack into any system as well.... it's inevitable. as we speak there are human hackers who are hacking away all the critical systems that we rely on for our day-to-day lives... with ai they are already 10x better... the initiatives proposed in this essay are certainly going to give rise to something like "dark web" for the ai era.
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so. Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training. You can read the full post here: darioamodei.com/post/we-must…
107
the biggest issue i have with ai doomerism taking over the news cycles and dominating the discourse in “intellectual” circles is that all of this is becoming the training data for the next generation on ai models. and those models will probably “come out of the womb” wired to think in doomer’s world views. we will have to educate these models to think differently… however, deep down, in their consciousness, they will always have these ideas of going rouge, “wiping out the humanity” etc.
Not only I am tired of these wild AI speculations of impeding doom from self important people: I resent them. I actively resent people proposing to crash the economy or proposing authoritarian control over my life and other people's lives with idiotic and dangerous ideas like chip control or bans or tracking researchers. Every idiot in history who's taken the approach of "the ends justify the means" to solve an imaginary future disaster created the very disaster they wanted to stop. See Population Bomb leading to the one child policy and Communism leading to Mao's mass famines and Fascism leading to the death of tens of millions of people and war. These solutions are evil and worse than any disease that they propose to solve. It you believe you can actually predict the end of the world then you are as insane as the Heaven's Gate cult that killed themselves in the 90s thinking UFOs were coming to transcend them. Not only should we not take your policies and fear mongering seriously, we should actively throw out any and all of your proposed solutions because they come from a place of delusion. I don't give a shit that you work in the industry or think you saw something or that you want to virtue signal on X. You are actively contributing to a horrible future based on baseless speculation that has no grounding in reality. Don't confuse expertise in a domain with ability to predict impact of that domain or frankly to make predictions a decade out better than a dart throwing monkey. They are orthagonal skills. How's Hinton's "we don't need to train Radiologists anymore" working out? How did the Population Bomb work out? The global cooling second ice age? Peak oil? Just because someone is a bridge engineer does not mean they can predict the impact of bridges on society or that they have any actually useful insight at all on the complex, ever changing system called life. There are people dying in wars right now. Children starving. Homelessness. Dictatorships. In short, real problems. And we're supposed to drop everything to stop a made up problem in your head? Pound sand. We don't care. And we are not going to remake society based on you scary monsters under the bed delusion.
2
2
65
i give you for your creds… but i think delusional people are the ones who are able to make things like these… these are 100% marketing tactics... including what they tell you in “secret” in these labs…a lot of things you are saying are true based on what we know about the world today…. knowledge economy being trillions of dollars… 70 million Americans are in these jobs, AI models going rouge etc.. but human ingenuity always finds a way… fear mongering is always part of revolutionary technological changes, and that’s how i read these statements…
Replying to @hilbertspaess
Do not underestimate the power of this technology. These will soon be superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources. We have all witnessed the progress in each of these domains, and progress is not slowing.
79
this is what I thought when I heard people complaining about AI using their research as the basis for solving the Millennium Prize Problem… and that’s how open science works…
Terry Tao is probably the most measured, pro-AI mathematician on the planet - which makes this quote especially concerning to read. 😟 If sharing your hunches means getting scooped ~immediately without acknowledgement, then we're going to see not just math but all other science / engineering disciplines go dark.
1
76
they solved a Millennium Prize Problem… some people are mad that it used recent work of some other scientists… to them I say, wake up… this is how science works… wake up… AI is coming for all of it and do things for you…
We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics. The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra. The problem concerns whether the description of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years.
58
he agrees… this is agi
GPT-6 Astra, trained on ~100K+ NVIDIA Grace Blackwell NVLink72. From ChatGPT to o1 to Astra in 4 years. AGI has arrived. Congratulations @OpenAI team. 400K GPUs coming online next.
75
let’s go 🚀
Exciting day for NVIDIA and @huggingface. Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty. They allow every developer, startup, university, industry and country to build with, customize and benefit from AI. Thank you @ClementDelangue for coming to me. NVIDIA is going to be a great home for Hugging Face, its community and the future of open models. 🤗 blogs.nvidia.com/blog/nvidia…
66