Justin Schroeder

The worst engineers I know write code by hand.

Genuinely excited to be speaking at Cloudflare Connect. The speaker list is absolutely absurd. Who else will I meet there? www.cloudflare.com/connect/

22 Sept

I can't believe this is a one-shot, Opus 5.5 just slaughtered model zen garden.

I’ve put a ton of tokens through Jev now, and...I have thoughts. 1. The big labs lost the script TypeSafe’s intro video says, “We’re building prod, not God” — and can I just say? Hell yeah.

I would even be willing to pay some kind of recurring license fee to have a great closed model running on my own hardware, just so I can reduce the TTD (Time To Decision) as much as possible. Imagine running a model like this at 60 Hz, or even 120 Hz.

At that point, we’re looking at a new kind of logic gate — we can’t even imagine what that will be like. In summary, this one really is a game changer for everyone building AI products, and, assuming cost and latency are further improved…it’ll be a game changer for everyone doing anything.

The good news is Diogo said on a recent AMA that they have substantial margin at their current pricing and have already considered dropping the cost further. I hope it's a lot further.

4. Latency is a real limiter I really hope TypeSafe AI works with an edge network provider to improve the co-location of models and reduce network latency. Ultimately, these models really need to run on-device.

To be clear, at current pricing, the model will still be successful and well integrated. It will be able to replace a number of tasks that were already being performed (poorly) by LLMs. However, to unlock industrial-scale demand (Jevons paradox), I believe it needs to be roughly 10x cheaper.

Currently, Jev is $0.042/M input, which is actually 7x more expensive than a cache read on DeepSeek V4.1 Flash ($0.006). If our industrial application requires decisions at 10 Hz and we provide only 10k of context, that would cost $0.0042/second to operate, or $15/hour, $600/week, etc.

These models will unlock AI utility for entire industries: robotics, trucking, aviation, meteorology, finance, defense, retail…even your smart coffee mug is going to use this.

3. It’s still too expensive Jev is really, really cheap compared to LLMs. But you don’t use it like an LLM. In my full self-driving example, or any robotics example, you’ll be making decisions several times per second continuously. The #1 value proposition of a model like this is speed and cost.

2. Ultra-smart classifiers were genuinely a missing primitive. I don’t know how this got missed. Diogo calls it “machine-native intelligence,” and that’s really what it is.

I cannot tell you how much bending and twisting I've had to do with LLMs to get them to act like a classifier when they just weren’t. I’m absolutely shocked this is the first time this “type” of model has come to market.

I cannot tell you how refreshing this is. It takes a sincere form of humility to believe that you, the creator of a technology, will not be the best at deploying the technology into the marketplace.

The big labs, especially Anthropic, have proven not to have this humility…and now everyone hates AI. Thanks, Dario. I guess what I’m saying is: Diogo for president.

@typesafeai built something for engineers like me, so I can build products for people like your mom (she says hi btw). They’re not trying to create “Machines of Loving Grace”; they’re trying to build things that let OTHER HUMAN BEINGS create new types of products.

The proof is in the pudding. When ChatGPT shipped, we were all blown away by…ChatGPT. When Jev shipped, we were all blown away by what everyone was making with it.

I rebuilt Tesla Full Self Driving with Jev in less than an hour. This model is a total unlock.

Tesla Full Self Driving (Jev demo) is now open source and you can try it online yourself! jevpilot.standardagents.ai 👈 try it github.com/standardagen... 👈 keep it

16 Sept

i spoke to a human being today that still writes code by hand. i didn't know that species still existed. was like looking at a living fossil.

21 Aug

We're open-sourcing ssh-clipboard. Finally, copy and paste between machines from a shared clipboard. - Copy/paste text, images, and even files. - macOS, X11, Wayland - Uses native clipboard - End-to-end encrypted (peer-to-peer over ssh) - Tailscale detection - MIT license

ssh-clipboard.standardagents.ai

21 Aug

ssh-clipboard.standardagents.aissh-clipboard — copy/paste, everywherePeer-to-peer clipboard sync over SSH. Copy anything on one machine, paste on another. No relay, no account, no new ports.

modelzengarden.comModel Zen GardenA long-horizon LLM benchmark. Explore two gardens, vote, view results.

12 Aug

If you haven't tried out modelzengarden.com its a pretty cool way to visually "see" the different models.

6 Aug

We’re announcing Standard Code: a new coding agent with unlimited use, like a cell phone plan. Each line has no 5-hour, weekly, or rate limit and powers one coding session. More lines = more parallel sessions. How is unlimited use possible? 👇

Standard Code runs in the cloud. Install a tiny daemon on any machine(s), then create new projects from the web, native Mac app, or the CLI coding agent. The expensive agent work is done in the cloud, so even a tiny VPS works great. Close your laptop and walk away.

standardcode.aiStandard Code | Always online. Unlimited usage. One flat price.Standard Code is a cloud-based coding agent with unlimited usage, built and operated by Standard Agents. $5 for your first week, then $49/month per line.

6 Aug

Standard Code is open for early access signups. ➡️ standardcode.ai

For Unlimited One, each parallel session requires purchasing a line of service. Just like a cell phone bill, it’s a predictable expense and won’t leave you out of tokens at the end of the week.

6 Aug

Sama One lets you bring your Codex sub to Standard Code, unlocking infinite parallel lines. It sips your weekly OpenAI usage because it, too, is a heavily optimized, curated collection of domain-specific agents built on OpenAI’s latest models.

In Standard Code you don’t pick models, you pick Agents. At launch, two agents are included: - Unlimited One - Sama One (for OpenAI subs) Each is a carefully curated, Pareto-optimized coding agent that uses many models across the various domains in coding.

2 July

Fable is back. just letting y'all know since things move a bit slower over here.

11 May

Faster than React, Vue, Svelte, and Angular. < 5kb and no build step.

2 May

Any internet friends in SF want to IRL?

vibebench.standardagents.aiVibeBench - The refreshingly honest AI benchmarkA manifesto and membership registry for experienced engineers building the coding benchmark that actually matters.

28 Apr

We’re announcing: VibeBench, a new benchmark for what actually matters — how models feel when used on real work by experienced software engineers. But, we need your help. Here’s how it works: 1. An initial cohort of 1000 qualified software engineers (join: vibebench.standardagents.ai) 2. ...

28 Apr

2. Groups of 250 evaluate new models for 2 days on real work. 3. Participants subjectively rank the model relative to other models they have experience with. 4. On day 4 a report is released with objective results derived from the subjective tests. How can you help: ...

vibebench.standardagents.aivibebench.standardagents.ai

28 Apr

1. We all need this benchmark to exist, but for it to become reality, we need an initial cohort of 1000 qualified software engineers. If that’s you, please join! vibebench.standardagents.ai 2. Repost this! 2. Share with everyone on your team. Lets make this a reality for all of us ❤️

27 Apr

We've made a huge mistake. We've forgotten decades of engineering wisdom. Composition over inheritance.

22 Apr

Mythos is sooooooooooo good. If only I had it, all my projects would be secure...

19 Apr

Starting doubt the Mythos narrative. If they have a God model then why: - Is Opus 4.7 so bad - Is Claude Code so buggy - Is Claude Desktop unfinished - Is the API down all the time