New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
my 3090
https://crof.ai/
Just taking api from apps , like codex, antigravity, kiro
It's the low end inferrence
Please share endpoint
https://kendell.dev/blog/crofaifalse/
what about cheaperinference.com? can anyone run an inspection like kendell one? @entrailz
I've done a fair bit of testing, the only two (with exception) I'd trust is inferhub and surplusintelligence.ai. If using inferhub, I would not select codebuddy, I can't confirm if it's rerouted but it's illogical the price offering.
yeah I saw it, that explains a few things lol
Was about to post this
https://www.reddit.com/r/opencode/comments/1wgdi37/crofai_cheapest_inference_provider_in_the_world
if you want to rent, i usually just use modal. hardly go beyond the monthly $30 free credits, but even then the pricing is ok
happy with 3090, running qwen3.8 flash next with some modifications. it's good enough for multi-subagent runs
impressive setup! i'll check modal, thx @ScreenReader
trying out https://deepinfra.com/ for now, will try out commandcode / opencode zen
deepinfra is usually available via openrouter (although they add markup). Opper.io is EU based, they seem to have a lower markup.
@entailz @steeler any luck on cheaperinference.com? i signed up, they give $10 free credits when you add $5, do you know if the free credits expire?
I sent a question to them about it and they locked my account so not looking good...
i agree, just use 9router or omniroute, a gpt plus can provide up to 100m tokens per week for me
Never tried - how do you sell there? If its all done via oauth to actual providers, it could be legit
Never tried - how do you sell there? If its all done via oauth to actual providers, it could be legit
skeptical of new providers after the Crof.ai fiasco, when requests were routed to cheaper models.
For light workloads I'd go pay-as-you-go rather than monthly plans, otherwise you're paying for idle time. Keep an eye on cold-start latency too, some budget providers are surprisingly slow on first call. Test with your actual prompt before committing, that's the only real way to tell.