Claude is beyond cooked

@lord.luke5

3.1K followers · usually 4.3K views a post

Views

110.1K

Likes

4.5K

Comments

334

Shares

376

SoundTikTok sound · Beethoven's "Moonlight"(871109)

Promoted product

DeepSeek V4-Pro 0813

Better AI results for less. Weekly model breakdowns, live builds, and honest feedback from people who actually post and ship.

Why it worked

The video leverages a trending topic (AI model performance and cost) and presents a clear, concise comparison with a surprising value proposition (DeepSeek's low cost). The direct criticism of a major player like Claude also generates engagement.

Summary

The video discusses the performance and pricing of various AI language models, highlighting DeepSeek V4-Pro 0813 as a cost-effective option compared to Fable 5. The creator suggests that Anthropic's Claude model is performing poorly and is too expensive.

Structure

1Introduction of AI model benchmarking
2Comparison of Fable 5 and DeepSeek V4-Pro 0813 pricing and performance
3Criticism of Anthropic's Claude model
4Call to action to join a course on 'Model Maxxing'

Original caption

Open sources winning right now and it is just crazy to think about #deepseek #china #claude

More from @lord.luke5

Product placement

DeepSeek V4-Pro 0813 ACTS - it is a large language model that is being benchmarked and compared to other models. Its performance and price are discussed, and it is implied to be a competitor in the AI space. Removing it would change the entire point of the video. Fable 5 ACTS - it is a large language model that is being benchmarked and compared to other models. Its price is discussed, and it is implied to be a competitor in the AI space. Removing it would change the entire point of the video. Opus 4.8 ACTS - it is a large language model that is being benchmarked and compared to other models. Its price is discussed, and it is implied to be a competitor in the AI space. Removing it would change the entire point of the video. V4-Flash ACTS - it is a large language model that is being benchmarked and compared to other models. Its price is discussed, and it is implied to be a competitor in the AI space. Removing it would change the entire point of the video. V4-Pro Preview ACTS - it is a large language model that is being benchmarked and compared to other models. Its price is discussed, and it is implied to be a competitor in the AI space. Removing it would change the entire point of the video. Claude ACTS - it is a large language model that is being benchmarked and compared to other models. The video's title states it is "beyond cooked", implying its performance is being discussed and criticized. Removing it would change the entire point of the video. Skool APPEARS - it is a platform where the creator is promoting a course. Removing it would not change the core content of the video, which is about AI models.

Call to action

Does opus 5 suck if yes join 👇

On-screen text

Claude is beyond cooked Terminal-Bench 2.1 Latest reported scores - price per 1M tokens (in/out) 88.0 87.9 85.0 82.7 72.1 Fable 5 $10 / $50 $0.435 / $0.87 V4-Pro 0813 Opus 4.8 $5 / $20 V4-Flash $0.14 / $0.28 V4-Pro Preview $0.435 / $0.87

Transcript

Skimpy has a big issue. Fable is only the smallest, the slightest bit better than DeepSeek. V4 Pro. And it's under a freaking dollar. Dude, Anthropic, you got to release Sonnet 5.5 and have it be insanely cheaply because you're so cooked.