Apollo @ApollonVisual

Tracking AI & Bitcoin Trends With Caduceus: 🔗 https://t.co/xPOhLouCXA 🔗https://t.co/1PLN1el5Kv 🔗 https://t.co/DtPggMjW9C @caduceusvisuals Joined September 2021

Tweets

4K
Followers

252
Following

199
Likes

6K

Mistral AI @MistralAI

3 hours ago

We’ve raised €1.7B to accelerate technological progress with AI! This Series C funding round, led by @ASMLcompany, fuels Mistral AI scientific research to keep pushing the frontier of AI to tackle the most critical technological challenges faced by strategic industries.

48 146 1K 70K 66

Qwen @Alibaba_Qwen

19 hours ago

🎙️ Meet Qwen3-ASR — the all-in-one speech recognition model! ✅ High-accuracy EN/CN + 9 more languages: ar, de, en, es, fr, it, ja, ko, pt, ru, zh ✅ Auto language detection ✅ Songs? Raps? Voice with BGM? No problem. <8% WER ✅ Works in noise, low quality, far-field ✅ Custom…

71 280 2K 175K 786

Download Image

Apollo @ApollonVisual

3 hours ago

The GenAI Divide STATE OF AI IN BUSINESS 2025 report was rather interesting. ictworks.org/wp-content/upl… While those who are heavily involved in AI suspected this (..."despite a massive influx of $30-40 billion in enterprise investment, a staggering 95% of GenAI projects are…

0 0 0 9 0

Download Image

Adina Yakup @AdinaYakup

3 hours ago

Baidu just released a reasoning model 🔥 ERNIE-4.5-21B-A3B-Thinking huggingface.co/baidu/ERNIE-4.… ✨ Small MoE - Apache 2.0 ✨ 128K context length for deep reasoning ✨ Efficient tool usage capabilities

4 25 148 4K 57

Rohan Paul @rohanpaul_ai

23 hours ago

💰 Google’s custom chips tensor processing units, or TPUs, are being seen as the strongest alternative to Nvidia’s GPUs, Research Analysts say if Google ever spun off this business along with its DeepMind lab, the combined unit could be worth $900 billion, of course they do…

39 123 1K 92K 303

Download Image

Tetsuo @tetsuoai

6 hours ago

X's Recommendation Algorithm Analysis ===================================== Used Grok Code Fast to get a quick breakdown of X's recommendation system. What Makes a post Go Viral =========================== tldr: Engagement prediction trumps everything. Post content that…

59 67 578 49K 637

Download Video

Michael Saylor @saylor

23 hours ago

Strategy has acquired 1,955 BTC for ~$217.4 million at ~$111,196 per bitcoin and has achieved BTC Yield of 25.8% YTD 2025. As of 9/7/2025, we hodl 638,460 $BTC acquired for ~$47.17 billion at ~$73,880 per bitcoin. $MSTR $STRC $STRK $STRF $STRD strategy.com/press/strategy…

1K 2K 15K 757K 266

Adam Livingston @AdamBLiv

4 days ago

WHY I AM MORE BULLISH THAN EVER ON MSTR AND METAPLANET The negative sentiment on here is HOGWASH. Compounding Bitcoin with intelligent leverage over a long time horizon = VALHALLA. Don't believe me? Let's run the math. 36 MINUTE DEEP DIVE:

158 299 2K 180K 635

Download Video

Apollo @ApollonVisual

a day ago

That was an interesting read: monostate.ai/blog/hallucina… OpenAI released a paper explaining why language models hallucinate. The core claim: standard training and accuracy-only evals reward guessing over admitting uncertainty. Hallucinations are framed as statistical errors…

0 0 0 17 0

Download Image

NIK @ns123abc

2 days ago

🚨 BREAKING: ASML invested $1.3 billion in @MistralAI at $14 billion valuation > ASML led Mistral's Series C > invested $1.3 billion out of $1.7 billion > becomes largest shareholder

69 202 2K 381K 297

Download Image

Apollo @ApollonVisual

a day ago

AI drama never ends it seems. U.S. senators folded the GAIN AI Act into the FY2026 defense bill. rswebsols.com/news/nvidia-ch… It would give American buyers first call on top AI GPUs and tighten export licenses using performance thresholds. Nvidia argues the bill fixes a problem…

0 0 0 44 0

Download Image

NIK @ns123abc

2 days ago

BREAKING: XAI COOKING ITS OWN AI INFERENCE CHIP > secured major ASIC order from Broadcom > inference only > target mass production 2027 > xAI actively hiring custom silicon engineers > Elon targeting 50 million H100 equivalent in 5 years

63 109 1K 111K 210

Download Image

Apollo @ApollonVisual

a day ago

Fujitsu unveiled a Takane-based reconstruction method that shrinks LLMs with 1-bit quantization. global.fujitsu/-/media/Projec… It reports a 94% cut in memory, 89% accuracy retention, and 3× faster inference. The goal: run capable models on edge gear like smartphones and factory…

0 0 0 20 0

Download Image

Luiza Jarovsky, PhD @LuizaJarovsky

2 days ago

Ladies and gentlemen, the generative AI industry:

81 634 3K 286K 3K

Download Image

Rohan Paul @rohanpaul_ai

2 days ago

📐 Wow. GPT-5 actually pushed forward the mathematical frontier with a result that had not appeared in the literature. Here, GPT-5 found out a real unsolved math problem by going beyond existing results and giving a new, precise answer that mathematicians did not have before.…

19 49 367 24K 161

Download Image

Neuralink @neuralink

3 days ago

Moments after implantation, our device can provide visibility into real-time neural activity. This video shows the gradual increase in activity as one of our recent participants wakes up post-procedure 🌅🧠

1K 2K 11K 3.9M 734

Download Video

Ethan Mollick @emollick

3 days ago

Paper from OpenAI says hallucinations are less a problem with LLMs themselves & more an issue with training on tests that only reward right answers. That encourages guessing rather than saying “I don’t know” If this is true, there is a straightforward path for more reliable AI.

136 259 2K 298K 757

Download Image

Andrew Curran @AndrewCurran_

4 days ago

Numbers for the Anthropic class action settlement: - $3000 per class work - $1.5 billion total settlement - LibGen and PiLiMi datasets to be destroyed - largest publicly reported copyright recovery in history

Garrison Lovely @GarrisonLovely

4 days ago

5 3 72 69K 20

Download Image

21 38 366 84K 114

Sam Altman @sama

4 days ago

thanks andrej! do you care more about it getting smarter or faster?

Andrej Karpathy @karpathy

4 days ago

thanks andrej! do you care more about it getting smarter or faster?

432 921 13K 2.4M 3K

945 327 7K 1.1M 759

Tibor Blaho @btibor91

4 days ago

OpenAI published "Why Language Models Hallucinate", explaining the root causes of AI hallucinations and proposing solutions to reduce them - Language models hallucinate because standard training and evaluation procedures reward guessing over acknowledging uncertainty, with most…