RussianCow

5 ms·

1.
by RussianCow 3d ago
Which doesn't matter if they're still not profitable.
2.
by RussianCow 13d ago
I don't see any kind of input cache discount listed on your pricing page. Do you offer that, or is all input priced the same?
3.
by RussianCow 13d ago
The issue is that all input (including context) counts towards that limit. So 10 requests with 50k of context will blow through the limit, even if little to no output was generated, which is incredibly easy to do with agentic workloads.
4.
by RussianCow 14d ago
This is pretty terrible advice when there are dozens of AI inference providers out there serving great models with significantly more cost effectiveness than you'd get from buying your own hardware.
5.
by RussianCow 14d ago
I don't trust any company with their word on anything. Luckily, privacy policies are legally binding.
6.
by RussianCow 17d ago
Presumably the number that OpenRouter shows is averaged across all requests.
7.
by RussianCow 17d ago
Most providers do what's called "prefix caching", where each turn in a session is cached such that sending new messages with the exact same "prefix" (set of previous messages) gives you the cache read price on that
8.
by RussianCow 17d ago
They're not "docking points", they're calculating it in the most straightforward way. If I start a session and the majority of requests are sent to Provider A, and my last request gets routed to Provider B, I have a 0% c
9.
by RussianCow 17d ago
But presumably everyone in your company/team is using Jira, so it's not an "ad" because it's a product already used internally. Claude is appending these links to all commits by default, whether or not others on t
10.
by RussianCow 17d ago
How else would you expect them to calculate it?
11.
by RussianCow 17d ago
This is very much NOT my experience in practice, even though it's how I would expect it to work. OpenRouter will happily bounce you between several providers (none of which have downtime) even within the same session. Requesting specif
12.
by RussianCow 19d ago
Can you give some examples of where it matters? I'm genuinely curious.
13.
by RussianCow 19d ago
That explains the difference, but what's the purpose?
14.
by RussianCow 19d ago
Interesting. I wonder where they're getting their data from then, because they list Z.ai under Singapore, but everything I'm finding says they're based in Beijing. Same with MiniMax.
15.
by RussianCow 19d ago
I don't think that's right, or if it is, OpenRouter has incorrect data. Several Chinese companies (headquartered in China) have Singapore listed as their region on OR. And some companies, like Alibaba Cloud, have multiple regions
16.
by RussianCow 19d ago
If you click on the provider name, the panel that pops up shows a "Region" value. Not every provider lists their region, however.
17.
by RussianCow 1mo ago
You might be joking, but a harness provides much more than just the prompts: at a minimum, it provides the system prompt and the built-in tools that the LLM can use, but it can also provide things like subagent management, custom compaction
18.
by RussianCow 1mo ago
I think it also helps that it's basically the default platform for any software that AI writes, unless you tell it otherwise. And JavaScript is one of the most widely used and well known languages in the world, so there's that, to
19.
by RussianCow 1mo ago
> A difference in degree But a pretty huge difference. It's the difference between a few individuals throwing waste into a river vs a corporation systemically polluting it. Surely one is more of a problem than the other?
20.
by RussianCow 1mo ago
Someone brings this up on just about every AI-related thread. I think it's nonsense. Nobody wants to maintain a fork of any remotely complex software, not even with AI. And in a corporate setting, nobody wants to use your custom fork;
21.
by RussianCow 1mo ago
That likely isn't as relevant for on-device iPhone usage as it is for Real Work™. I won't notice the difference between 50tps and 1000tps when asking Siri a question.
22.
by RussianCow 1mo ago
There's no reason you couldn't present a GUI over an SSH connection. Whether the agent is running locally or remotely is orthogonal to the type of interface used.
23.
by RussianCow 2mo ago
I'm not confusing anything since you can't use custom harnesses with your Claude subscription—you have to use Claude Code. So as far as the Claude sub is concerned, the models and the harness are coupled.
24.
by RussianCow 2mo ago
They're not as good. I've tried all the options available a couple months ago, and none of them feel nearly as polished as Cursor. For one thing, Cursor Tab is just far better than any other "next edit" prediction I&#x27
25.
by RussianCow 2mo ago
> the results are much better than clowning on bloated CC. I won't argue with the cost effectiveness, but the results are very much not better. Opus and Fable are in a different league than DS4 Flash. Even GPT Terra, which I reall
26.
by RussianCow 2mo ago
The problem is that companies tend to exaggerate their own scale and think they need k8s and dedicated infra people when they could get by with a handful of beefy VMs or dedicated servers.
27.
by RussianCow 2mo ago
I haven't used them but aren't they basically modernized Heroku? What's different?
28.
by RussianCow 2mo ago
This seems like an arbitrary line to draw. Should your photo viewing software also not provide editing tools? If it's useful, why not?
29.
by RussianCow 2mo ago
It depends. For something high stakes or inherently complex, sure, you don't want to have to clean up the agent's mess afterwards. But for many tasks like building web UIs, the difference in output quality is going to be small eno
30.
by RussianCow 2mo ago
Once you've used a model that runs at hundreds of TPS, it's hard to go back. Everything completes so quickly that you can iterate without breaking out of flow state. My biggest gripe with slow (<50tps) LLMs is that I've lo

More ›