@lord.luke5: Did Claude Opus 5.5 get nerfed? Honestly, it's pretty hard to tell. Opus 5.5 is Anthropic's newest AI model, and people started saying it got quietly made worse just days after launch. NerfBench, a tracker that tests it every day, had it at 103.8% of launch power on October 1 and 94.2% on October 2. That's a drop, but still inside their normal range. My take: the better these models get, the harder a nerf is to spot, because for the most part the output still looks great. Have you noticed a difference? 👀 Claude Opus 5.5 nerfed, Claude Opus 5.5, did Claude get nerfed, Claude nerf, is Claude getting dumber, Claude Opus, Claude AI, Anthropic, NerfBench, AI model nerf, AI benchmarks, AI news

2026-10-04 01:57:08

2026-10-03 22:16:28

2026-10-03 23:17:46

2026-10-04 07:03:34

2026-10-04 17:32:55

2026-10-04 08:16:51

2026-10-04 00:58:07

2026-10-04 17:13:24

2026-10-03 22:42:57

2026-10-04 01:07:12

2026-10-03 22:21:19

2026-10-03 23:08:37

2026-10-03 23:39:44

2026-10-03 22:31:10

To see more videos from user @lord.luke5, please go to the Tikwm homepage.

Unc Luke

Open In TikTok:

Region: US

Saturday 03 October 2026 22:11:11 GMT

Music

Download

Comments

PhjClips :

Astra lowkey been cooking since everyone canceled and went to claude

dirt :

They did the same with Astra :/ I bet it’s because they hype it up at the start for money or whatever then make it bad for whatever reason

. :

AI isn't getting that much better anymore so they nerf the previous model so the new model feels a bigger leap forward than it actually is

Invincible8o8 :

better than astra still

gamesbygianni :

It’s working amazing for my game so. Scrapped my last project because the AI just wasn’t there yet, it works great when used with real human output and creativity

MonoluminAI :

Not sure about that… it loosing focus now…. But I cannot tell for sure

Kozi :

They just got bad context

nate.builds.mov :

they always quantize the model. thats the point. nobody says it publicly but they do.

H :

I personally think they can’t make more powerful models so they make the current models weak and rerelease an improved version of the same model.

PopsTheWise :

It did. it did get nerfed for like a day. now literally anything I throw at it is only 2.4k tokens

Eabusham2 :

Last time this happened with Astra OpenAI addmitted to changing some model settings for a select group of people so who knows

Mehdi.øufc1 :

i’m still waiting on qwen 4 moe it’s gonna be crazy i’m currently using qwen 3.6 35b a3b moe but on codex it keeps cutting out mid work idk why

Zakk :

Can you talk about Gemini 3.8 routing to argon some dude said it with proof on Reddit

rustygooners :

that's why use Chinese

Other Videos

About

Legal