solarkraft

That’s an impressive image, but what a mistake it was to click the second link (on mobile without an ad blocker). I wouldn’t send it to anyone I respect ...

How is Cornell’s ornithology lab so strong? They also produce Merlin Bird ID (https://merlin.allaboutbirds.org/) and work on the underlying BirdNET (https://birdnet.cornell.edu/).

“all about birds” indeed!

Yes. You can also do it outside of it just fine.

Larger screen support coming soon:

Checkpoint 3: Board v2.0 Release (Supporters)

Status: Check in around September 2026!

Expand the platform beyond the original 4.26" display and demonstrate the full capability of the optional module system.

(…) 4.26" (same as initial prototype), 7.5", 3.97"

https://github.com/iandchasse/de-link/blob/main/markdown/ROA...

I also have issues with OpenCode. It induces a few facepalms and bafflement in a few key areas (“why the hell would you do it like that”).

Yet I still use it because they have made very sensible decisions in other key areas! The server/client split makes all the sense and the harness generally just works and has all the features that I want. They’ve done a great job with the core user experience.

Because of the clean client/server split, I can currently fully focus my energy on building a client that works just the way I want. When I’m done with that, I might tackle fixing the server to work the way I want.

If someone here is interested in maintaining a version of OpenCode that takes itself more seriously, I might be on board with contributing. The last time I looked such a project didn’t really exist with significant traction, unfortunately.

Despite my occasional facepalms when learning about how the stuff is made, I remain super grateful for OpenCode. It’s the only real fully-featured open alternative to the proprietary (I’ll include Codex in that) harnesses and has given me tremendous value for free.

I can understand both modes. I mostly use transcription as input for my AI assistant and there I find it very useful to be able to check my input and just repeat myself in case something wasn’t fully captured. When using Apple’s transcription feature built into iOS and macOS, I also really like being able to edit everything right while the dictation is still active.

Does it support streaming? I find that this is the #1 thing missing from almost all implementations.

Oh, I like this! I’ve been looking into locally hosting a transcription API server and came away feeling pretty close to the problem statement. The things most frequently lacking were streaming support (which I’m so glad this has!) and the support for special words to boost during recognition (which I guess there’s some hope they might add???).

I am saying that people here seem to be appealing to Microsoft as an authority that should be interested in stopping this, perhaps because they are morally superior or concerned about their reputation. They are not.

Yes! Extra apps suck.

They own Typescript? I wasn’t aware that they control the organization, but that ought to be easy enough to fork. NPM is a bigger one, but also not too huge. Azure is only used by people who already have Microsoft/Windows buy-in.

You’re acting like Microsoft aren’t pushing malware themselves.

Maybe they optimized it really hard for small models. That would be impressive, but hardly a good selling argument. I want a harness that can do both and OpenCode looks just fine for that purpose (and it also has everything else I need).

To me this looks like another case of bundling things that shouldn’t be bundled (the harness with the UI) making both worse off because you can’t individually focus on each component. They’ve done this before, bundling a decent UI with decent inference for IMO no good reason, combining the downsides of each instead of letting people mix+match.

Yes. I don’t see it either. It looks like a competent app (converging on the same principles as others) but what they are advertising as differentiators simply isn’t relevant to its purpose.

It’s a harness + UI with really confusing messaging. Made for open models ... why specifically? Zero data retention ... well duh, it’s a client????

What a respectful view of the audience. Too bad this approach was replicated what feels like approximately 0 times after it.

https://youtube.com/watch?v=kl6rsi7BEtk

I have not tried Wero, but I have tried the german predecessor (paydirekt/giropay). It was ... not much less smooth than PayPal.

But I’ve only ever seen a single vendor offering it. 0 for Wero so far.

Because it's a CRDT behind a local web server

I think that’s the secret to stability. I see too many projects that try to embed the harness into their UI. I think that’s completely wrong because it leaves no interface for changing or reasoning about the system.

The harness needs a lot of love and it should be as small and as possible (at least somewhat). I dream of a world in which we have some standard interfaces here, including for the UI.

Thanks for the read. It seems to confirm that resource limits are an important factor for terminal benchmarks:

The extra resources enable the agent to try approaches that only work with generous allocations, such as pulling in large dependencies, spawning expensive subprocesses, and running memory-intensive test suites.

An agent that writes lean, efficient code very fast will do well under tight constraints. An agent that brute-forces solutions with heavyweight tools will do well under generous ones. Both are legitimate things to test, but collapsing them into a single score without specifying the resource configuration makes the differences—and real-world generalizability—hard to interpret.

So changing the resource limits changes the benchmark. Yet their score table claims their score to be for Terminal-Bench 2.1, not Terminal-Bench 2.1 with raised limits.

They cheated again: https://news.ycombinator.com/item?id=48847019

Congratulations!

I also love exploring ideas like this. The problem that made me stop using Gemini is that it would always try really hard against all my prompting to ask a stupid question at the end of its response that would completely derail my train of thought.

I love the UX of voice mode. I’ve always hated voice models.

Gemini’s being particularly egregious (always ending in some deranged question, can not reliably be prompted away) prompted me to build my own client for my real harness that simply does STT -> model -> TTS (both being independently useful).

I guess I see some value in a model responding quickly and with more nuance, but it’s not much. I can wait for it to finish. I’d much rather have it be actually useful. I’m not looking for a digital friend.

The delegation feature lets me see some value in a voice model for orchestration type features. But in either case, I don’t really like (or understand why others would like) talking to a model with different features and quirks just because I’m using a different medium to communicate over.

It basically already is. You can’t use the search logged out and my agent can’t fetch issue comments.

Why a Python library instead of a completions API proxy?

Maybe it doesn’t matter, but I tend to tell people when I have the rare good support experience.

Alternative framing:

Instacart sells data to Papa Johns for better pizza ad timing.

Papa Johns isn’t somehow able to predict this based on their own data (which is the impression I initially got from the headline), they are buying it from Instacart.

Name and shame.

Deepseek regularly does this for me, at least when actively exploring something (multi-turn). It massively varies its reasoning budget based on the task.

My impression is that they use OpenRouter, so market price among multiple inference providers.

build an image from scratch with the packages you want, composed in the way you want them

That’s exactly what this does, elegantly