381–390 of 433 posts
Earlier quoted context omitted.
Every single one of those sentences is highly dubious. Cyber-defenders would be pretty jazzed about having easier access to Mythos-class models. Cyber-defense is easier with better tools.
Cyber defenders already have access to Mythos class models at most companies with the biggest user facing products and services. They are using them extensively (including on OSS deps) Handing every skiddie and nation state and APT and hacker group access to Mythos does not help cyber defenders Even if you don't think Mythos is a big deal: At a certain point models become smart enough as to be dangerous, and you don'…
Luckily, Anthropic has no such monopoly, so the point is sorta moot
Earlier quoted context omitted.
Check out LibRedirect or Predirect (MV3), it automatically redirects youtube, X, etc links to privacy-respecting frontends. https://news.ycombinator.com/item?id=44344246 107 comments, 1 year ago.
The LibRedirect link in that post is 404
Earlier quoted context omitted.
> single line addition to prompt to use subagents proactively. This misses an important detail. In Claude Code [1], ultracode suggests the agent create a JavaScript code to deterministically orchestrate sub agents. This is different from just having the main agent launch sub agents and (non-deterministically) manage them. The resulting workflow is called “dynamic” because CC creates this orchestration script dynamica…
Can you explain what you find useful about ultracode? I've become wary of agent swarms since the early days and now just prefer to have a single agent spin for hours at time. Parallelism never got me anywhere worthwhile.
Earlier quoted context omitted.
> single line addition to prompt to use subagents proactively. This misses an important detail. In Claude Code [1], ultracode suggests the agent create a JavaScript code to deterministically orchestrate sub agents. This is different from just having the main agent launch sub agents and (non-deterministically) manage them. The resulting workflow is called “dynamic” because CC creates this orchestration script dynamica…
Can you explain what you find useful about ultracode? I've become wary of agent swarms since the early days and now just prefer to have a single agent spin for hours at time. Parallelism never got me anywhere worthwhile.
- parallelism: "if it makes sense, use ultracode to do this thoroughly" Claude decides if this is the type of task that can benefit from a dynamic workflow. For example where there is a large number of similar independent sub-tasks, it can choose to set up a dynamic workflow, and it even sets up verifiers/tests etc to validate its work. In some cases it pushes back and decides not to set up a dynamic workflow and instead works on the task directly.
- verification loops: "Set up a dynamic workflow with adversarial checkers, and iterate until they are all green, up to some sensible limit". Similarly, I can include code-simplification steps and documentation-update steps.
These patterns help me leave it unattended for longer periods of time and return to a reasonable assurance of correct code.
Earlier quoted context omitted.
Are you really making the case that teaching a person how to work is equivalent to prompting an AI?
Yes. Tell me why it’s different without using circular reasoning?
Teaching modifies the learner. Prompting doesn't modify the model. It provides additional context that influences a single inference. A person who has learned something can apply it years later without being reminded. An LLM generally cannot unless the knowledge is incorporated into the model itself or provided again.
Earlier quoted context omitted.
> single line addition to prompt to use subagents proactively. This misses an important detail. In Claude Code [1], ultracode suggests the agent create a JavaScript code to deterministically orchestrate sub agents. This is different from just having the main agent launch sub agents and (non-deterministically) manage them. The resulting workflow is called “dynamic” because CC creates this orchestration script dynamica…
Can you explain what you find useful about ultracode? I've become wary of agent swarms since the early days and now just prefer to have a single agent spin for hours at time. Parallelism never got me anywhere worthwhile.
Earlier quoted context omitted.
oh here we go again saying everything in china is done by the state. it's not a democracy but they're not the soviet union either. if that were true, why don't they have just one model initiative instead of several?
Oh yeah that's a good point. In that case, this Chinese hedge fund is choosing to sink billions of dollars in R&D alongside vast opportunity cost in order to create models and release them for free because umm... they're just that nice! No wonder you're confused about DeepSeek when you have a fairly obvious explanation provided to you, and your response is "it's unrealistic to think the Chinese Communist Party is beh…
to you, it's plainly obvious that a non american, non-european company engaging in open source and releasing research means they are state owned and seek to destroy the US when this line of thinking is exactly the result of the red scare.
Earlier quoted context omitted.
But there must be a ton of generic questions that people ask. Stuff like "What's the capital of country X?" - it's probably at least 10% of queries. Memories, custom instructions etc would invalidate them, but if you can return the answers basically free it's probably worth it.
How is that cheaper? You now need to have a database of millions of possibly gigabyte sized rows. Also, transformers have quadratic complexity, so short queries cost practically nothing. The only optimization that makes sense is per user prefix caching, because you are often sending the same system prompt over and over again or are continuing a conversation.
Earlier quoted context omitted.
As of writing, https://openrouter.ai/rankings shows only Opus, Sonnet, and GPT 5.5 in the top twenty by usage (Opus and Sonnet both have two slots on the 4.7/4.8 and 4.5/5 splits respectively, read that how you will). I don't know how much business OpenRouter does in Europe (they have some GDPR text in some settings pages I think) but it's zero in China. You might also consider the countless companies that do nothing…
Misleading first sentence, as most use OpenAI and Anthropic directly via their own LLM not through OpenRouter. But you're right that lots of companies use open weight and Chinese models, not sure how the parent got to their conclusion.