Hacker News: ishurand4

New comment by ishurand4 in "Claude Fable 5"

ishurand4 — Wed, 10 Jun 2026 22:52:07 +0000

The only one I see that thinks it is claude other than claude itself is the GLM series.

New comment by ishurand4 in "Claude Fable 5"

ishurand4 — Wed, 10 Jun 2026 22:45:17 +0000

Its a quadratic graph. It starts low but not that capable, gets better and more expensive, and then the time comes in which the capability needed is not the ones of the frontier models and then the price goes down on the companies who host the models that the capability is "good enough"

New comment by ishurand4 in "Claude Fable 5"

ishurand4 — Wed, 10 Jun 2026 21:37:25 +0000

A myth, not a model that actually exists

New comment by ishurand4 in "Claude Fable 5"

ishurand4 — Tue, 09 Jun 2026 21:56:34 +0000

A system "card" made mostly by the model itself.

New comment by ishurand4 in "Claude Fable 5"

ishurand4 — Tue, 09 Jun 2026 21:55:27 +0000

Thats why it is a mythos model

New comment by ishurand4 in "Claude Fable 5"

ishurand4 — Tue, 09 Jun 2026 21:55:10 +0000

Well, for me at least, I pay more for input (Up to 1M per prompt) than output (usually max 4k-8k)

New comment by ishurand4 in "Claude Fable 5"

ishurand4 — Tue, 09 Jun 2026 19:00:34 +0000

Yeah, it is also known as Claude Mythos 5

New comment by ishurand4 in "Claude Opus 4.8"

ishurand4 — Thu, 28 May 2026 20:00:18 +0000

What's the token usage at?

New comment by ishurand4 in "Claude Opus 4.8"

ishurand4 — Thu, 28 May 2026 19:31:50 +0000

The numbers they show don't matter. "On multi-round coreference/context recall tests (often cited as MRCR or long-text retrieval benchmarks), Opus 4.7 reportedly dropped from roughly 78.3% down to 32.2% compared to Opus 4.6.", but what did anthropic do? They just stopped showing the benchmark altogether and then just show the cherry top ones that got improved on.

New comment by ishurand4 in "Claude Opus 4.8"

ishurand4 — Thu, 28 May 2026 19:28:25 +0000

And anyway, with quantum, there will be no need for frontier companies as you might be able to even run a 1T param model on a consumer quantum computer.

New comment by ishurand4 in "Claude Opus 4.8"

ishurand4 — Thu, 28 May 2026 19:25:40 +0000

They just showed the benchmarks it improved on but it regressed on so much more, such as the MCRR benchmark: "On multi-round coreference/context recall tests (often cited as MRCR or long-text retrieval benchmarks), Opus 4.7 reportedly dropped from roughly 78.3% down to 32.2% compared to Opus 4.6."

New comment by ishurand4 in "Stack Overflow’s forum is dead but the company’s still kicking"

ishurand4 — Tue, 26 May 2026 19:00:20 +0000

Stack Overflow and Reddit are still getting threads. And as AI gets smarter, the questions will also expand.

New comment by ishurand4 in "Ask HN: Shouldn't Google need to give a public statement about Railway incident?"

ishurand4 — Fri, 22 May 2026 19:17:58 +0000

But then they have alot more services since msft bought them, making higher chances on goes down