mark_l_watson

Yeah, I mostly agree with you on both points. 1. Tech execs are all in for making money. A small tangent: my wife and I used to enjoy the All In Podcast, but now those four guys mostly lie (my opinion) in ways to profit themselves and their rich friends - really out of touch, and now they are kind-of boring. Used to be a fun podcast. 2. I am a super techie, retired now (I have 55 patents, written many books on AI, many great jobs): I am a little shocked at how most non-tech people I talk with don’t like AI: some because of energy use/data centers forced on unwilling communities, many fear for their or their children's or grandchildren’s jobs, etc.

No, just keep the usual tax/finacial/health data on my devices.

I consider Anthropic's Mythros security bug finder mostly marketing, but other things worry me that there might be a global hack contagion: for example, a few months ago I saw in the news that an executive at a US security company was caught selling information to a hacking group.

Except for disabled Javascript compilation possibly slowing down web sites, not getting some attachments in messages, and some graphics not showing up on some web sites, having Lockdown mode set doesn't seem to affect anything I do. For dev I use VPSs with ssh set for ensuring SSH agent forwarding is strictly disabled, as are reverse tunnels.

It seems like doing little things like this make sense because it is such a tiny hassle to be a little safer.

Not sure how much it helps, but I just run all my Apple devices in "Lockdown mode", don't install apps (use Safari), and try to mostly use Safari in private sandboxed mode.

yes, and Claude is available on Google AntiGravity with some paid accounts.

I can guess: I am 3 weeks into a 4 week Ultra subscription and the amount of Claude Opus and Gemini Pro tokens that they give you on the subscription is very generous - I feel like I have been gorging on tokens, tidying up 25 years of my open source projects. When my one month subscription runs out I will miss it.

Maybe this will be a market success? My non-tech brother loves all the AI features on his Pixel phone.

Apple has a good story with Handoff to connect Apple devices. Google might have some success selling a Googlebook + Pixel phone combo.

Cons: Google cancels products that some people love.

Pluses: I think there is a real chance that Anthropic and OpenAI tank financially and Google might win consumer AI, with enterprise AI shared between Microsoft, Amazon, and Google.

I am in the US but my daily chat AI is Proton's Lumo+. Adequate, good features, not too expensive. I am near the end of a 1 month experiment using Google's $250/month Ultra AI plan and I wanted to try something different when the Ultra plan ends. I have tried Mistral's Vibe command line coding agent and considered it, but decided on a one month OpenCode + Deepseek v4 experiment.

I understand why Europeans might want to go all in on their own tech stacks, but it might be more strategic to just not get locked in to specific providers. Maybe a mix of European, US, and Asian tech - with a good plan for easy migration.

That looks cool. If they support binary releases on macOS Apple Silicon I will try it out.

I wrote one major project in UCSD Pascal on Apple II 45 years ago: my Honnibo Warrior Go Playing program. For its time, a fantastic dev environment.

I agree. I worked in Singapore for a short while and I was so very impressed by the people and the government.

Singapore is a great place. It is breathtaking to see a government govern in the interests of its people. I live in the USA and we have two institutions, the democratic party and the republican party, that do little except cater to special interests.

I used to be a huge fan of the All In Pod Cast before the four hosts started spending too much time going on and on about their $2000 bottles of booze, and too much bragging in general, and too much politics. Have a good life guys, but tone down the bragging, that is boring. Seriously, previously their business and tech podcast was the best. This article is pretty interesting.

thanks for the link! Interesting paper and little-coder looks purpose built for my own local model agentic experiments.

When you run ollama serve, make sure you override the context size to about 32K. Also, I give the model a useful short README.md on the code it is writing or modifying, and a Makefile with useful targets for the agent to use. I usually use Claude Code with qwen3.6

I also go outside for fresh air while I wait for a session to run.

The politics and economics of Musk throwing some support towards Anthropic is interesting (samma is probably pissed).

But, if you will pardon a little rant: I hate the idea of subscription inference plans and also 'dumping' by subsidizing non-profitable products. Inferencing should be pay as you go and dumping illegal.

I like Google’s business model more than the other frontier model providers: sustainable. One thing I don’t like with Gemini Ultra is no visibility into token use or what the cost would be. I have been planning on letting my Ultra subscription expire and go with OpenCode with a fast inference provider to get this visibility, but this discussion thread gave me the idea of also trying the paid APIs with AntiGravity instead of a subscription. When I sit down to do a specific task I want accurate token usage and $$ data as I work.

Yes! Sometimes when models get something wrong with less widely used programming languages, I like to just cancel the current inference, fix something myself, then tell the harness/model that I fixed the current problem, and to move on.

I have been experimenting with using Claude Code with both the qwen3.6 31B MOE and 28B dense models. Yesterday the 31B model once got confused on refactoring some Prolog code and took a very long time to get it right. Functionality for coding or refactoring Python or TypeScript is usually good. but runs slowly on my 32B MacMini.

Ollama has initial support for bf16 MTP Gemma 4 https://ollama.com/library/gemma4:31b-coding-mtp-bf16 but I have to wait for a smaller model.

I understand why people get excited by having the strongest AI to play/work with but economic factors of inference really count also.

I was in their Pro plan for about a year, now on Ultra, and I am planning on downgrading to the cheap $8/month plan and just using OpenCode with a good inference provider because of one thing: I like seeing my token and cost $ usage data in real time. I know this sounds a little crazy, but I like visibility into what resources I am using.

I think subscription plans are a little bit evil.

Th said, Ultra with the initial half price deal is awesome: all the Opus tokens I need in AntiGravity.

I am trying to wrap my head around this: if I remove Chrome Browser, will I reclaim the disk space for this model? Thanks in advance.

I love TUIs because so much of my dev career was ssh/mosh/tmux to remote servers and working where the fast network bandwidth and computational crunch lived.

I was vibe coding a layer on top of textual last week that allows me to plug in both interactive and information display Python functions into a TUI grid. Really simple stuff, with textual doing the heavy lifting.

That would have been very useful to me, before I retired! That said, I only run the Hermes Agent on leased VPSs and PyInfra might be a cool and easy to access Hermes - I need to think about that.

Awesome. I fly very small remote controlled airplanes and a tiny drone. I am a bit nearsighted and I fly my airplanes and drone relatively close to where I stand (10 to 75 feet). A friend flys a few large RCs, but I don’t think that size matters for having fun. I live in the mountains in Central Arizona, and I like to hit the flying fields just as the sun is starting to rise in the morning - beautiful time to fly.

BTW, 60 years ago my father and I used to spend a lot of time building our RCs. To be honest, now I buy incredibly inexpensive planes from China.

I agree. I have rather constrained use cases for LLMs and the agentic harnesses that I use with them.

I try one or two of my use cases with new models or harnesses, make my own often subjective judgements, and largely ignore benchmarks.

Blogging and writing in general are a business, or feed other tech adjacent businesses, and a lot of writing about evals is attention getting - nothing wrong with that but there is a lot of noise.

Chiasmus Looks very cool. I might have a use for it because I like to use LLM harnesses to explore code. Thanks.

Oh, thanks!

I bought a 32G MacMini over two years ago and it has been great for experimenting with local models, and now is even useful for local coding (at a slow speed!) with models supporting large context sizes.

With the current extreme RAM shortage I deeply regret not buying a 64G MacMini a few months ago.

I bet a zillion people feel the same way.

The shortages are caused by memory manufacturers in Asia having supply chain problems, and the pending strike at Samsung, right?

Will this require a paid plan? That is, could Warp, when modified, work with a local model on Ollama, no charge?

A word of warning: I just installed OpenWarp from source, but it looks like it will not let me use my own provider without signing up for an $20/month account -- just like the original Warp

I very much wish the OpenWarp folks would have made this clear on their README.md file.

it is usually fast, but if gemini-cli or any other coding agent is sluggish I quit using it for a while.

I like the idea of Mistral, but the last time I evaluated Mistral Vibe it was really nice for $15/month but not as effective as Gemini Plus with AntiGravity and gemini-cli. I am currently running Gemini Ultra on a 3 month 'special deal' and AntiGravity with Opus 4.7 tokens is pretty much fantastic.

That said, when I stop spending money on Gemini Ultra, I will give Mistral Vibe another 1-month test.

I like the entire business model and vibe of Mistral so much more than OpenAI/Anthropic/Google but I also have stuff to get done. I am curious if Mistral Vibe for $15/month is a stable business model (i.e., can they make a profit).