jerrygenser

$0.10/mm input vs. $0.042/mm input. Both free output.

github action implements other providers by essentially a request to your runner. so if actions go down, the action supervising your different provider would likely not run so you're going to have an outage anyway

ipod of 2026 ships q4 2027.. is that a typo on the shipping date?

I'm not sure if they don't refer to it as "stealing" but they refer to "distillation attacks"

I wonder if they are signalling that if they can do this for training, then they can create an style agent swarm to hack anyone with 380k concurrent agents.

that is the the "data used to improve the model" when signing up for the subscription plans

Of a lot of the AI contrarian language out there in blog posts, I actually thought this was not so bad

> The company also emphasized that the model is faster and more efficient than its predecessor, GPT-5.6 Sol, on a variety of tasks. For example, OpenAI said that Astra achieved a higher score using fewer output tokens, a common unit of measurement for AI tasks, on a key cybersecurity test called ExploitGym.

Releasing open source models can drive revenue away from them US AI LABS. This can help china win by depriving those labs of revenue for further investment in winning the long term race.

Python but also add

Uv -> OpenAI

Perhaps you were served from someone else's cache

This uses separate git worktrees. If we have a local dev setup involving multiple docker services, is there a recommended solution for managing those envs? I didn't see.

so if you pick max-5x or mx-20x than pro... are you getting 0 now?

It's also good as an @explore sub-agent that greps the directory for files.

Acquisition of noam shazeer to supercharge their Gemini flagship model line I think made a bigger impact.

To make an argument it was Kevin Hou, then we would need to see Antigravity their new IDE being key. I think the crown jewel are the Gemini models.

A common use of asyncio is a server framework like fastapi that schedules tasks. I used such a framework for a while before realizing that I needed to create_task for within-task concurrency.

Basedpyright is not rust but it's a fork of pyright with added features that are otherwise locked in vscode

You still have to write the resolver for graphql. I've seen. N+1 with graphql if you don't actually use data loader+batch pattern OR if you use it incorrectly.

Corporations are picking the side that's in power. If team blue is in power they would pick blue. Corporations are (usually) not moral or inherently politically motivated other than to the extent of optimizing short term shareholder value.

Re the overnight that's why some providers are offering there are batch tier jobs that are 50% off which return over up to 12 or 24 hours for non-interactive use cases.

Why do you suspect it was vibe coded? There are 2 substantive files that are each less than 100 lines...

Also the readme doesn't have the usual emojis for every bullet point.

You could remove venture and it would still be accurate

They can lock in by subsidizing the price of you use their tool, while making the default price larger for wrappers. This can draw people from the wrapper that can support multiple models to the specific CLI that supports the proprietary model.

I think this is like mass data labelers, including in other countries or other low cost areas.

What is your future revenue model?

I also got into materia magica for a few months about a year ago. They are still actively developing it! Get into the discord

Yes it does. And it will fail if there is a combination that is not resolvable.

You do not write requirements by hand. You write requirements.in and uv pip-compile to requirements.txt

Is there any security risk to use as project with an embedded .wasm binary? Is there a build process for this that we can reproduce, or is there a way to bring our own if it's a generally available asset?

Referring to this file: https://github.com/embedpdf/embed-pdf-viewer/blob/main/packa...