It's probably polarizing and I honestly don't know why I felt the need to rant about it. /shrug

I just use dumb and fast models now. I'm more engaged. I think that the higher the quality of the model, the more you tend to vibe with it, and then the more hallucinations you then miss. I'm not sure which is more productive, but I definitely burn out faster the more I vibe. At some point you're spending your time on forums, discord, or youtube instead of engaged with what you're building. Or you yak shave about your tooling and end up creating the 600th multi-agent gastown harness and blowing thousands of dollars on tokens to create it only to discover it's too expense to actually use.

FWIW I've noticed that agents are pretty good at writing "High-Level Rust" for most basic applications, which gives you pretty great performance (orders of magnitude faster than RoR), great deployment, probably great security if steered by a senior, and pretty great maintenance again if originally steered by a senior. I feel like this is a not-so-secret secret.

I personally won't be using dynamic languages for anything but toy scripts now. (well, except JS, which is hard to avoid with the massive size of WASM bundles)

P.S. I assume Go is still great as well, but IMO Go no longer has an identity. What are they going for anyways? Garbage collected rust? "How to invent a perfect niche and then throw it all away in 21 days". /rant

I don’t trust real estate either. Seems like anything with a middleman is setting record levels of grift right now. I don’t trust the industries reports. No one is regulating or checking numbers.

I shifted into more bonds. Probably a bit early, but I’m not a pro. I’ve just lost trust in the market which no longer seems tied to reality or at least my limited understanding of reality. Staying in it just feels yolo atm.

I basically have it load up a bunch of relevant context and give it small chunks of work in the same session over time (not like a fire and forget subagent). It's working fairly well. Bonus is I still feel like I'm part of the process instead of watching youtube videos while Opus / GPT vibe code a bunch of slop.

Codex 5.5 extra high currently feels a good amount smarter than either 4.6 or 4.7 Opus. I only just started using it about a week ago, so maybe that's a recent development and then OpenAI will eventually lobotomize their model or throttle etc.

What I dislike about frontier models is how opaque and incentivized the businesses are about tweaking their services. Anthropic definitely does some shady throttling. I have zero trust for Altman and he's BS AGI claims. And Google makes it non-obvious that you can't turn Gemini training off on even their highest tier personal plan. There's a lot of shady and dishonest behaviour, probably because they are all overhyped and heavily subsidized to win the race. I don't mind at all paying more than I currently am for these services, but I don't trust any of these frontier model companies, and so I'm cheering for open models.

Right now I'm using Codex for planning and DeepSeek V4 Flash [1m] for implementation. It's quite fast. Quite possible / likely that OpenAI will make significant changes that kill this workflow for w/e reason... at which point I will probably move to full open weight models.

Something like a proxy that intercepts and depending on the source, is intelligent enough to examine the package for age. That would be cool. Already sounds like a cloud product you could sell.

I’m not in love with bankers running the country either, but give us another option.

I'm very pro union in the private sector.

I'm coming around on this too. deepseek-v4-flash is impressive.

It would be nice if this was a bit more obvious and clear too.

----

alias clauded='ANTHROPIC_BASE_URL=https://api.deepseek.com/anthropic ANTHROPIC_AUTH_TOKEN=$DEEPSEEK_API_KEY ANTHROPIC_MODEL=deepseek-v4-pro[1m] ANTHROPIC_DEFAULT_OPUS_MODEL=deepseek-v4-pro[1m] ANTHROPIC_DEFAULT_SONNET_MODEL=deepseek-v4-pro[1m] ANTHROPIC_DEFAULT_HAIKU_MODEL=deepseek-v4-flash CLAUDE_CODE_SUBAGENT_MODEL=deepseek-v4-flash CLAUDE_CODE_EFFORT_LEVEL=max claude'

----

I doubt that the opus, sonnet, and haiku model args actually matter if you want to omit them.

I run this on a VPS that has no other credentials or project access so I can give it the skip permissions arg.

This is what I’ve been using for non-confidential projects for about a week now (soon after v4 came out). I honestly can’t tell the difference, but I’m not doing anything crazy with it either.

Worth noting that I don’t think DeepSeek‘s API lets you opt out of training. Once this is up on other providers though… (OpenRouter is just proxying to DeepSeek atm)