Talk:Claude Mythos

The repository may be https://github.com/anthropics/claude-code

though there is also OpenMythos a theoretical reconstruction not affiliated with, endorsed by, or connected to Anthropic or any of their proprietary systems.

That repository may be at https://github.com/kyegomez/OpenMythos/

Should the claude-code repository be added to this page or is Claude Mythos not based on part of that claude-code code but shares that claude name? I do not yet know a repository for the large language model/data/code that works with any of this type of software.

Other Cody (talk) 18:26, 22 May 2026 (UTC)

Claude Code is a coding agent. OpenMythos is an effort to recreate Mythos. Neither should be on the page. elijahpepe@wikipedia (he/him) 18:41, 22 May 2026 (UTC)

Summarize Timeline Top Qs Fact Check

@Maxeto0910: Instead of edit warring with two different editors over this, please explain on the talk page why a PR announcement blog post belongs in the external links section when we already have a link to the official website for the product. I dispute that this clearly belong here as an exception to Wikipedia's norms for these kinds of links. As a primary source, it could be cited (again) as a citation for specific, non-controversial details. Otherwise, this is just corporate bloat without context.

As I assume we all already know, Wikipedia explicitly isn't intended to be a directory, nor is it a platform for PR. Grayfell (talk) 01:27, 11 June 2026 (UTC)

[...] with two different editors over this [...]

Well, you two had different rationales for not including the link, neither of which I think applies.

[...] please explain on the talk page why a PR announcement blog post belongs in the external links section when we already have a link to the official website for the product

I pretty much implicitly did so in my edit summary by quoting from the guideline. I'd argue that the page provides unique and valuable information that is not covered by the other external link. If we had to provide only one link, I'd even argue that it makes more sense to only include the latter.

I dispute that this clearly belong here as an exception to Wikipedia's norms for these kinds of links.

Okay. Why?

As a primary source, it could be cited (again) as a citation for specific, non-controversial details.

If there aren't any third-party sources available for verifying a claim, sure.

Otherwise, this is just corporate bloat without context.

How is it without context? It provides specific information about both models, including examples of use cases, benchmarks, results, feedback, videos, links, and more. Not wanting to include a useful link simply because it is from the company behind the product seems pretty biased. Maxeto0910 (talk) 01:44, 11 June 2026 (UTC)

Yes, I am biased against using Wikipedia for promotional purposes. This is about a work of PR in a 'see also' section. See also sections do not include any context. The encyclopedic value of these examples of uses, 'benchmarks' etc. should be demonstrated by reliable sources, not the company, and not your personal endorsement. We cannot trust this company to neutrally present these bits of information. Since Wikipedia still isn't a platform for public relations, I am again removing this as spam. Grayfell (talk) 19:40, 27 June 2026 (UTC)

Yes, I am biased against using Wikipedia for promotional purposes.
Yes, the information is presented in a promotional way, no one has ever denied this. And yes, despite this, I still think it provides valuable information. Removing it simply because it is promotional is a very weak argument IMO and seems like an instance of following dogmata as an end in itself, rather than evaluating whether WP:ELMIN applies or not. Like I already wrote in an edit summary, the guideline says: "More than one official link should be provided only when the additional links provide the reader with significant unique content and are not prominently linked from other official websites." I'd argue this is clearly the case here.
We cannot trust this company to neutrally present these bits of information.
Yes. And avoiding a link to a website providing valuable information simply because we don't have confidence in our readers understanding how PR works is something I consider patronizing. By that logic, we shouldn't link the official website as well. Also, every Wikipedia article about a product is inevitably in itself already a weak form of PR in the sense that it lowers the friction for people to learn and talk about its capabilities.
The encyclopedic value of these examples of uses, 'benchmarks' etc. should be demonstrated by reliable sources, not the company, and not your personal endorsement.
Correct. That's why we don't include them in the article but as an external link. Maxeto0910 (talk) 21:54, 1 July 2026 (UTC)

Providing this link to readers tells them that this information will be specifically useful for understanding the topic. That's the basic purpose of such links. I dispute that such benchmarks are valid metrics at all, and much of the rest is just fluff. Perhaps the metrics are valid in some ways, or perhaps they are industry hype. You don't have to agree with me on that, but since two editors have reverted this addition, you do not have consensus, regardless of whether or not you accept that validity of our explanations. This is a dispute about editorial discretion. To put it another way, including this is a form of editorializing.

One way to address the issue, at least as I see it, would be to use a WP:IS to provide context in the body. Of course, this is what the article should be doing anyway. If this is impossible, that shows that the link isn't worth preserving here after all. Grayfell (talk) 23:36, 1 July 2026 (UTC)