Claude Fable 1.5 ist raus!

Hook: „A new Entropic model from CLD is out, and we're going to see how good it really is.“

Short·2:11 min·veröffentlicht 03.09.26 (vor 1 Monat)·erfasst 07.09.26

1,0×Faktor (Views / Median 1.330, n=31)

1.375Views

8Likes

2Kommentare

Talking Head + B-RollFallstudie / ProofHook: insider wissen90s-3minProduktion: mittelUntertitel: einwort karaokeCTA: kommentar keywordKonfidenz mittel

Warum es so läuft

Der Post liegt mit Faktor 1,00x genau im Median dieses Accounts, performt also weder über- noch unterdurchschnittlich. Der Hook ist Insider-Wissen über ein neues Modell, was zur Nische passt, aber offenbar keine zusätzliche Reichweite auslöst. Sichtbar sind Talking Head mit eingeschnittenen Screenshots der Benchmarks und ein Einwort-Karaoke-Untertitel, eine für den Kanal übliche Produktionsweise. Die geringen Like- und Kommentarzahlen bei 1.343 Views deuten auf ein kleines, aber stabiles Nischenpublikum hin.

Belege

Ausreißer-Faktor 1.00x, exakt Median
Hook nennt konkretes neues Modell (insider_wissen)
Talking Head mit Screenshot-Einschnitten (Benchmarks)
Einwort-Karaoke-Untertitel im Thumbnail sichtbar
Nur 8 Likes und 2 Kommentare bei 1.343 Views

Hook

„A new Entropic model from CLD is out, and we're going to see how good it really is.“

Titel-/Hook-Muster

Produktname + Ausrufezeichen als Neuigkeit

Für Stefan / Heartbeat

Ein System Breakdown Format, in dem Stefan neue KI-Tools live im Vergleich zu bestehenden Prozessen testet, könnte ähnlich funktionieren, allerdings mit stärkerem Business-Bezug statt reinem Tool-Vergleich.

Thema

Vergleich neuer KI-Modelle Claude Opus

Sichtbar

Person im Talking-Head-Format vor heller Wand
Schwarzes T-Shirt
Einwort-Karaoke-Untertitel eingeblendet
Screenshots von Benchmark-Tabellen im Video
Bäume und Zaun im Hintergrund oben sichtbar

ki-toolsclaudemodellvergleichtech-reviewbenchmarkscreenshots

Thumbnail-Stil: gesicht emotion

Analyse 07.09., 06:57 · claude-sonnet-5

Verlauf

Messung	Alter	Views	Likes	Komm.	Label
07.09., 06:39	3.8 d	1.343	8	2	t0
14.09., 06:48	10.8 d	1.375	8	2	t7

t0 = erste Erfassung, t7 = etwa 7 Tage, t28 = etwa 28 Tage nach Veröffentlichung.

Notiz

Beschreibung

–

Transkript (apify, 335 Wörter)

anzeigen

A new Entropic model from CLD is out, and we're going to see how good it really is. And now you see, it's already showing up here for me. Let's try it out right away. How many retail orders are currently open and how long does it take for them all to be sent out based on historical data, only SNOs . Let's send it off with Fable 5.1 and right here with Opusit. As for the benchmarks here, you can see that the new model is about three points better, so around 5%, jumping from 33 to 66. The speed has also gone up from 54 to 66. So, a jump in speed, which looks very similar at first glance. Both create a table right away. Um, it's abbreviating the numbers for me here. It's not written out here yet, but I have to say, I actually find this one more detailed. It breaks it down for me more specifically historically. How long does it take? So, outstanding, around 110,000 items of note. So, at first glance, I have to say, because it's in bold here with Opos, I actually find this result better than the new one. My personal conclusion after just one prompt, but also after reading some reports this morning. It is smarter and more clever for really complex tasks. As a CEO, I don't even need that. I don't need an even better one to build something crazier. For me, the current factors that are important are speed. How fast do the models respond to me? Both are similarly fast, so not really faster, and above all: how precise is the output? Also, how many output tokens it has, so how long are the answers, and how quickly does it get to the point? And I have to say: when I factor all of those things in, I'm actually a big fan of Chat GBT/codex right now. If you want a Reel on that too, write it in the comments.