MegaLLM Discord Insights

not sure why, no model works for me due to a payment error, even though there are still some money on my balance

Model name ?

Gemini 3.1

You need paid credit to google models

So not sure why but I'm getting a error saying I'm out of credits when I haven't even used half of them yet

let me see

make a #channel and give me error

👍

I can no longer use Claude models like Sonnet or Opus due to a "payment error", even though they worked before and the site still lists them as available through promocode money. Is something wrong with the models?

there was some issue we are trying to fix meanwhile you can use other models

Alright, thanks for the answer

Claude isn't listed for me even though I bought credits and I'm getting the same error. Is this also a bug or am i doing something wrong?

open a ticket send receipt and account u have mail on

will fix

Done 🙂

I tried to top up $5, but I was actually charged $498.79. Could you please check this? On my card statement, it shows a charge of $498, but in Mega LLM, only $5 has been credited.

I haven’t used anything since making the payment. Is it possible to cancel the transaction?

WHAT!?

man, i'm about to recharge too $5 too, but i'm scared that it happens me the same

I paid through Stripe, but there was a payment error. When I went back to the Stripe payment page to check, the incorrect amount was still showing on the payment screen.

So that is a visual error?, and it won't really charge me what random price is?, i'm just afraid of lose a lot of money

for now, should i buy credits? just $5 i'm planning to buy.

The issue has been successfully resolved and I have received a refund. It appears to have been a Stripe-related error. When using a US VPN, the correct amount is charged.

Same here. Look like a joke. They give us "free" credits, but we can't use them. Things never change.

wrong input and output being shown in the website than actually what it is supported.

it was an VPN error but its resolved

Free credits + stability is what matters if u would have said with this gemini and claude will also work but broken than, you would’ve bashed us more

thanks for letting us know @Saurabh Pandey will get it fixed ASAP

There's a huge amount of supporter who can claim credits, which will make claude and gem have more down times since the load will be through the sky. So giving it to only paid users makes it stable

Making high value models available for free (i know you guys paid, I did as well but I charged to access models) Is not sustainable

I was an owner of something like an api, acted like a backend for a site, huge consumption for free is not stable at all

I was a paid user once before—and you know what? For a month and a half of my subscription, absolutely nothing worked.

so with credits you can still use a lot of models if u are having any issues in them let us know we will fix

This "other" things I can use in other places near for real free. Because I can't even test your Gemini or Claude for understood real price after this invisible discount, I never pay megallm for this again. This project finally dead for me now.

you need to understand something if we give everyone to test now it won’t be able to handle and its like when market will get stable we will get u for “testing” and if u want them for free we can’t help you should opt for free rather than better uptimes and assurance

Thanks for your time with MegaLLM if u need anything or have feedbacks do lmk

getting an error using claude opus 4.6: undefined is not an object (evaluating 't.rpm')

is it in a specific request or contineous

Contineous

Is it fixed now? Can you check

yes it is now, thank you very much

I keep getting PROXY ERROR 503: {"error":{"code":500,"message":"Resource overloaded, please try again after some time","type":"overloaded_error"}} (unk)

For glm 5 and deepseek v3.2

its getting fixed, will udpate asap

sorry for the inconvenience

Np

Grok 4.1 fast in both no reasoning and reasoning it's not working. Just an infinite loading.

working on it will update asap its solved

Yeah... Grok 4.1 fast isn't working again.

will fix it asap

Emm, I cannot buy credits using Stripe method. Anyone else?

what issue is it showing @Saurabh Pandey will look into it and get it fixed asap

It has been fixed , sorry for the inconveniences caused , u can try once .

The MegaLLM promotional model has a bug — it rejects tool_choice: {"type":"auto"} (which the AI SDK sends by default with tools), but works fine if tool_choice is omitted.

i was using opencode btw, and opus told me this was issue

1

@😊 hey u can dm me with the issue

opus 4.8 overladed people, we wait patiently:20251124_1232_SmilingHaloEmoji_r:

almost every good functioning model is overloaded :20251124_1232_SmilingHaloEmoji_r:

it has been fixed

all fixed

Pretty much every model I have tried returns resource overloaded and I even upgraded my subscription to get more priority access

Plz help 😭

yea I also experience the same thing as well who the hell spamming this 😂

@Tamish @Saurabh Pandey @Priyanshu. Good morning, is there any news on this?

I still can't access the products i use daily that haven't yet been accomodated to allow for the switching of models easily, because I set them up with some MegaLLM models to begin with. I am aware that this set up is a flaw in some of my own projects so I am working on making sure it easier to swap out models from different API keys. But for pipelines and workflows that need a set model or family of models to function, I can't use any of them because all of the fallbacks I have set up are still within MegaLLM and it would appear as though a MAJORITY of the models I have set as fallbacks in case an initial model, goes out or down, aren't working either. But at this rate, what is the point of paying for and using MegaLLM when I end up having to use credits from the original sources half the time because more than half the models that i use in MegaLLM aren't wotking?

To me, it feels like I should have just saved half the money I invested in the current plan for MegaLLM back into the original providers or to OpenRouter or something. So that way I could use MegaLLM models the half of the time that they are working, and then i could actually use the other half of the money to buy credits from a provider whose models actually work consistently and who have fallbacks that work consistently.

are you sending temprature to the models?

@t.azul.ai@Tio can you guys create a ticket and get me some more details about the issue? so that i can help you better

getting this request error: Internal Server Error {"success":false,"error":"Failed to process LLM request","detail":"Unable to connect. Is the computer able to access the url?","errorType":"Error"}

from Opus 4.8

will be fixed asap will ping when fixed

this is a network error from your side

Hm am not sure where to post this . But this is the issuse i having now : I buy the premium plan today using UPI (razorpay). Everything working properly. Then after half an hour I cancel the automatic payment. Now it's saying : "payment required". @Tamish

open ticket send mail will resolve also don’t cancel subs like this😭 u can cancel them one day before billing cycle

I usually forget about automatic payment...so i usually cancel them before hand. Which mail id i send mail to? Ticket no is 400

send mail u had acc on in ticket

I have barely been able to use my subscription, almost everything returns { "error": { "message": "Resource overloaded, please try again after some time", "type": "overloaded_error" } }

Every model or just claude models?

I just tried with gpt-5.4 mini, same error

Ah, models might just be down for the being at the moment

Been like this all day for me

More of the same today

@This guy farts try gpt 5.5

seemed to work for now, looks like anthropic is still busted for me this is all I see from sonnet when I prompt it with pi {} Tool not found {} Tool not found {} Tool not found {} Tool not found {} Tool not found

but yes, so far gpt-5.5 is working well

503 error with gemini 3.1 two days running. If it keeps going this way I'll just cancel my sub if it's so untable.

it will be fixed soon will notify u once done sorry for the inconvenience caused

Seems like Gemini-3.1-pro-preview is still down. Gemini-2.5-pro also seems to have gone down along with Claude-ppus-4.6.

PROXY ERROR 503: {"error":{"message":"Resource overloaded, please try again after some time","type":"overloaded_error"}} (unk) Opus 4.6

Still getting a whole lot of 503's

Which model? @This guy farts

Was gpt 5.5 and Gemini flash 3.5

Currently getting 503s with Gemini 3.1 pro preview and 2.5 pro.

PROXY ERROR 503: {"error":{"message":"Resource overloaded, please try again after some time","type":"overloaded_error"}} (unk) Opus 4.6 again.

do you guys have a bug for pricing with model mimo-v2.5-pro? somehow similar input and output with kimi-k2.6 but kimi cost 1/10 of mimo-v2.5-pro despite kimi model pricing is much more expensive actually not sure if mimo or kimi is the one bugged but cost should be similar given the input and output cost..

fixing it

Termux saying: it's provider 232 which gives the response. But usgae page say : it's provider 238? Is it bug?

glm 5.2, kimi 2.7, ds v4 pro all resource overloaded it was working 3 or 4 hours ago..

Opus 4.6 overloaded.

@Saurabh Pandey 503 errors again..please fix

on which model @huh? Me

Opus 4.6 and sonnet 4.6 was giving 503.. But now it's working...thanks for fix

I haven't logged into my MegaLLM account for a while and noticed that the bonus credits I was being given out had somehow disappeared, forcing me to re-register. Is this some kind of scam, or why did you delete my account?

Add this WTF!!!?

Any success yet?. @Demondralla

GLM 5.2 returns no providers available for this model

Yes

@Demondralla open ticket

How?

Get ticket = Error Give me instruction guide.

Any success yet @Demondralla

@Demondralla fixing it soon

Claude opus 4.8 won't use thinking no matter the parameter passed

I also receive this error whenever i try to log in with my google account.

I purposely deleted all of my API keys since I noticed some usage on my account that I did not recognize. I'm still seeing usage show up on my account today that happened after I deleted all of my keys. Also I'm seeing usage for a model that I know I've never used.

Any success yet?. @— Aventurine! @This guy farts

/redeem

Opened a ticket, haven't heard anything else yet

There is a huge bug with the RPM for paid plans. The documentation says that a PRO plan is supposed to get 300rpm but I've been capped at 20 for the past few weeks - months and it is insane. That is less than 10% of the suppsoed RPM I'm supposed to be getting.

Opus is down

Opus is Overload