First LONGGG post on X. Soooo, some quick intro:
Hey, I'm Jasper from Singapore๐ธ๐ฌ. (My nickname is aeongenosis)
Some facts about me:
- Beginner building startups/apps (more on this soon!)
- Songs genres I listen to and love: Country, Country-Pop, MandoPop
- For English songs, more of a Country/Country-Pop fan than Pop
- Love reading non fiction books (about anything.. have read STEM books to the arts!)
- For Fiction books, I only read Romance bookz :).
Personally, I will also be using this account for a variety of purposes:
1. Post about certain apps/startups I'm building + its progress.
2. Post long form articles on History, Philosophy and Geopolitics (on X and Substack).
3. Post about any independent studying of textbooks or MOOC courses on topics of my interests.
4. Post about the financial markets.
This is just a repost by him but they are not 100% wrong though. In essence, research comes from human intution. How can we allow AI to autonomously solve many of our research problems not just in the field of Mathematics?
I'm not against using AI for research but the sole reliance/usage of AI for research is certainly worrying IMO.
But again, I believe everyone and anyone who has an interests and will to learn and explore the field can now independently research a field aligning with their interests with the help of LLMs.
@TheH0n3stTrader but frankly, i think it's more than 3 years.
I'll say at least half a decade to ensure consistent profitability.
At most 2.5 years with a (helpful) mentor.
@TheH0n3stTrader nah, for a beginner, the worst periods of life are today and tomorrow. Afterwards its 100% winrate with 10x-100x capital and driving lambos and enjoying life.
/s
๐
I humbling disagree.
Apple already has Google.
Tbh, Amazon do not need their own model. Their focus is selling the shovels (aka the compute). IMO, that's sufficient. It's pointless for Amazon to venture into getting their own model from a biz POV.
Their core businesses are Cloud Computing and E-Commerce. Both of these do not demand their own models.
Furthermore, Apple should just focus on hardware and making their hardware user friendly (aka focus on design). This itself will drive their revenue.
Google being chromebook? ๐
Agree on the rest.
For general-use cases (other than programming):
Willing to bet more people will use Gemini in the next decade as compared to OpenAIs' or Anthropics' models if the cost of subscription for Claude/ChatGPT continuously increases.
Google has the pricing advantage as they can offer AI premium with their current stack. Also, lets not forget that Gemini flash is the fastest and smartest as compared to Luna or Haiku (not too sure on the lastest benchmarks).
I love how Google, a trillion dollar company, is now the underdogs among the major US labs.
This is soo much fun ๐
Edit: Haiku 5.5 (which was just released hours before this post and after Gemini 3.8-Flash) is doing quite well on cost wrt intelligence. Ignoring that, Google seems to be doing quite well imo.
Current AI Scene:
Anthropic is Apple - easy recommendation for everyone.
OpenAI is Windows - It's really good and most people know about this but in some cases, it will hinder your everyday performance.
Gemini is Chromebook - no one shoud use it apart from school kids so that the teacher can see a really bad work and immediately tell that Gemini was used.
Qwen, GLM, Kimi - These are Linux distros.. Some distros are better than others in some areas but they are all generally pretty good.
Mistral is TempleOS
@chrisbarber This looks very skewed towards US labs.. Most of the consensus are for US labs but all Chinese labs are either half-consensus or over-half.
Top crypto on my watchlist, in no particular order (NOT Financial Advise --> personally watching):
1. Ondo
2. Arbitrum
3. Raydium
4. PONS
5. ETH
For me to keep track too!
Meet Mistral Large 4, aka Le Chonk.
โข 1T parameters, natively multimodal. 49B active.
It is the best open weights model from US or Europe on aggregated benchmarks.
โข State-of-the-art on critical workloads, including cyber defense, manufacturing and finance and it surpasses closed frontier models on visual grounding.
โข Forged in Europe end-to-end and is deployable from Europe via our own Mistral Cloud infrastructure.
โข Available to all via API today. Working with cybersecurity partners privately.
Open weights release end of October.
From Closed-Source model races to Open-Source model race...
Which US startup will be next to release a new frontier western model.
The next playbook for AI startups:
Pivot to training OS models --> Benchmarkmaxx it --> Market is as a frontier OS model --> Hype it up --> Valuation rockets to Billions of dollars --> Raise Series Z --> Sell to some tech company which is interested in training an inhouse model--> Paydayyy!
Introducing Beam: a highly efficient agentic open model with 501B total parameters and 23B active.
- Frontier reasoning efficiency
- Advances the Western open frontier on coding & agentic tasks
- Trained end-to-end from scratch
Full weights release this month.
Learn more about Beam: https://t.co/c3Qx2cpM8G
We are making Markdown files work natively across Google Drive and Docs.
You can now preview .md files in Drive, open them in Docs, edit them, comment on them, and collaborate without converting the file into a Google Doc!