A complete walkthrough of the new Claude Sonnet 5 release FULL Claude Code bootcamp in the Vibe Coding Academy coming up: https://www.skool.com/vibe-coding-academy 2nd Youtube Channel: https://youtube.com/@AlexFinnLabsOfficial Sign up for my free newsletter: https://www.alexfinn.ai/subscribe Follow my X: https://x.com/AlexFinn Henry Intelligent Machines (my new startup): https://meethenry.ai My $300k/yr AI app: https://www.creatorbuddy.io/ Timestamps: 0:00 Intro 0:41 What Sonnet 5 is 3:41 Performance test 6:25 Sonnet 5 best practices 10:15 Fable 5
ADVERTISEMENT
GPT 5.5 medium benchmarks better and cheaper... I wish they made sonnet 5 cheaper!
Should be called Sonnet 4.8
You buy a 10k pc and a bunch of mac studios and say that it's for local AIs and you want personal data and now you say you are giving away your ID to Anthropic.
Kimi k2. 7 usually finds stuff that opus 4.8 ultra with fanning eub agents can't lol
FABLE BACK TOMORROW! FULL LIVE FABLE BOOTCAMP IN THE VIBE CODING ACADEMY!! JOIN NOW:
huh ??? Have you carefully read the pricing of Sonnet 5 and watched the intelligence? When working at high or xhigh you pay the same price as Opus 4.6 but without the intelligence or context length... it's totally not worth using sonnet 5 at this pricing setting. Better keep using Opus 4.8. I really have no idea why they released it at this pricepoint. And when yo uset Sonnet 5 on low, then it is dupmber then Haiku but at a higher price point..
no more coffee.....no more blow bro
Completely skipped over High and XHigh at 3:15 though lol. Wonder why
Not really “a fraction of the price” I mean sure it is cheaper than Opus 4.8 but like GLM 5.2 and other models are way cheaper than this so…. Idk.
Opus 4.6 came out a month and a half ago?
Paid subscriptions for different AIs slowly empty most individuals pocket 😂
Dude sonnet 5 costs more than opus with the tokenizer and extra burn
Rage bait surely
I think you need to learn to read a chart. Sonnet 5 is slightly MORE expensive than Op. 4.8 at the high and extra high levels while performing more poorly on each. This model makes no sense other than for the masses to not realize that they’re going to spend more money for worse output. Very disappointed in Anthropic re this.
Rubbish! GLM 5.2 kicks its arse, and is way cheaper (actually free on Z AI web).
How do you guys manage to fully assess a new model that fast?
It's not really back if it's not available to US allies...
Watch the price, tokeniser 1.3 AND its token burn heavy on the tasks. 10% saving including the discount vs opus 4.8 Disappointing devil in the details here
Its pretty terrible actually. Same pricing at Opus 4.8 if you take into account its tokenizer. This should be Sonnet 4.65.
Click bait and Shit i have used it