So I've been praising Qoder for a while now. Still is, the most consistent and best external AI platform I've used (and I've used ALOT)... But the past 2 weeks have made me question their motives...
For anyone not up to date, Qoder released a new model option, Sonus. Which is marketed as "SOTA". It cost 3.2x credit rate (while Qwen 3.8 Max was at 0.5x during day, 0.2x non-peak and Kimi K3 at 1.4x). So on release, as a 'give it a try', they offered free credits for Sonus, but it was a trap. 500 credits, at 3.2x rate, meant it lasted all of 2 minutes, it doesnt even write a line of code before it was out. Then 1k credits (half of what you get for 20 usd a month), also gone in 3 minutes, not a single line written... Then 2k credits and still not a single line written and gone in 4-5 minutes...
Now they raised the price to 8x... More than double the price, for a model that has no documentation, no param count, no benchmarks, nothing... Essentially, they could be marketing qwen flash at high thinking and nobody would know any different, because it'd cost you 100 usd just to get a single line of code out of it.
This IMO, is anti-consumer behavior, baiting users to try it, because they know they'll overshoot. Chances are, the model doesnt cost them 6x what max costs them... Let alone 16x! It's absurdly expensive to drain a user's credits the second they refresh a second too late. For perspective, at that pricing, you could likely get more work out of Claude Fable for the price... Yet they charge it on a model that has literally NO information?
This isnt acceptable behavior and while I love Qoder as a platform, this is anti-consumer behavior and I wont let it go unheard. If you use Qoder, please dont use Sonus, dont even 'try' it, because it's marketed as a guaranteed way to drain your credits. Whether that's because they're in the red, or because they named it Sonus (Sonnet), same way they named it Cantus (Claude), with no documentation, to try and market Qwen as the 'smart choice', it's not how you do business and it's honestly an insult to every user. When they marketed Max, they offered free calls, which were way more manageable and lasted 1000x longer. This... This is not it. This isnt marketing, this isnt promoting your new model, this isnt even business, this is daylight robbery on something that brings no value.
If you use Qoder, ignore Sonus and Cantus, use the qwen 3.8 flash promo to it's fullest, because that's the Qoder I know. Good value and fantastic promos! Not this BS marketing 'SOTA' while obliterating users' wallets... I shudder to think what happens when someones on pay as you go and goes to make a cup of coffee when 20 usd worth of credits lasted Sonus at less than half the current price, less than 5 minutes... Gods forbid you go to the bathroom too!
I dont think I've ever seen a clearer AI scam than this... And it's clear that their intention is to force users to use Qwen, that's why they marketed it as such and why they coincidently offer 3.8 flash for free in the meantime! Because they want to show 'for the price of Sonnet, or Claude, you could have Qwen all month long'... If that's the marketing you want to do, then post benchmarks, cost comparisons, genuine statistics, not cost users a month worth in 10 minutes...
Top comments (4)
Hahaha🤣, this is the classic playbook of big platforms. They get you hooked with free credits first, then start monetizing hard. I’m kind of used to it now. The key is to keep several alternatives handy and never put all your eggs in one basket.
Yeah, noticed that on all the platforms, just thought Qoder, being Qwen backed would be different (my mistake was trusting)... That being said, the naming... Sonus, Cantus... They're clearly jabs at Anthropic, Sonnet, Claude... Marketed as 'SOTA' that gets no work done for the price you get a month's worth done on qwen? It's a marketing ploy, that's why they wont release any info on it, because it's not an internal model, it's Anthropic's models, that are being used as a way to market Qwen! What pisses me off though, is that they do it in a way where if you're not actively monitoring it, refreshing every 10 seconds, it drains your entire month's quota in no time... But what they're missing, is that they're comparing pricing against API cost, vs Anthropic offers a token plan, that actually gets the same amount of work done as Qoder (if you dont use the free models)... It's clearly a marketing stunt to get people off of Claude, but the reason it pisses me off, is that they dont pause before it switches to draining your normal credits, beyond the free credits? That's just anti-consumer behavior. It's honestly the type of thing I'd expect from Anthropic, I wouldnt have guessed that Qoder would pull that stunt, given that they market it and Cantus as 'internal models'... But you cant tell me that it's not Anthropic models... They intentionally market it at an insane price, with thinking on max, so it runs out of credits before it writes a single line, so they can create a value moat for Qwen! When you spend enough time using a model, you start picking up on it's characteristics and this is definitely Sonnet... Only Anthropic's models can spend 20 minutes thinking before writing a single line (that's not a complement btw) and the wording... It's not a chinese native model, it's definitely a western model... And the pricing fits. 'SOTA' at 3.2x price (guess again), it's 8x... But when you actually look at it, Qwen 3.8 Max is at .5x (2 usd in, 6 usd out), but Sonnet 5 (2 usd in, 10 usd out) isnt THAT much more expensive? But put it on ubercode thinking level, it'll take this long and the 'thoughts' it produces is identical.
I honestly expected better from Qoder and it's saddening to think they're no different.
Hahaha. 🤣When engineering meets business operations, your worldview will shift a lot. From many projects I’ve worked on, a key priority has always been figuring out how to make users burn through their prepaid balance or credits as quickly as possible. That’s why I’m more excited about advances in local LLMs. Honestly, I really dislike sending data to the cloud; it never feels secure.ðŸ«
xd