how so? it literally says `api.openai.com`
how so? it literally says
api.openai.com
HOh… 
HYeah, that’s my bad
HMaybe a geoblock then?
HWait, geoblocks wouldn’t work on Workers IPs
Swe have users from 140+ countries and never had this issue so doubtful
HOr a ban on a specific CF account
HSince you should be able to see the incoming zone
Swhy not?
HAren’t they all anycasted?
HLike, I don’t think GeoIP functions on them
HGonna test later
Shmm not sure but I do know when we make requests to some services the data is often localized depending where the actual user is from
Sit would be really really nice if you could specify a proxy to use in a fetch
Sjust had to do an ugly workaround for this
Swould be incredible for things like sending traffic through whitelisted IPs
HAn HTTP-based proxy should still work, no?
SI could not find a good solution to use with a dedicated proxy server
Swell I did find a solution but it is to go through our scraping service which has a feature to specify your own proxy which is nice but it means going through yet another server lol
HNo, like the Hoppscotch Proxy
Swill create something with node-http-proxy or something later but I didn't have time
HEverything is a post request, with a JSON body
Soh
Swell this could have been nice to know about
Sam I blind or why can't I find documentation on how to use the proxy 
HI think it is only documented for use with Hoppscotch
HThough what I meant is that it shouldn’t be too hard to build a small Rust app that forwards requests for you
Syeah not hard we just have bigger priorities and this was to just get around one large host that decided to block all Workers
九Cloudflare’s AI is fantastic. I am testing the text-to-image model @cf/stabilityai/stable-diffusion-xl-base-1.0 and encountered some issues, so I would like to ask:
- Does it only support the parameters ‘prompt’ and ‘num_steps’, or are there other parameters available as well? For example, ‘negative_prompt’? I couldn’t find this information in the Cloudflare AI documentation.
- The official XL model supports inpainting mode and image-to-image mode. Are there any plans to support these?
Ahey, i have the same questions
九Also, I would like to ask if anyone has encountered the following situation: today it seems that openai.com has blocked many IP ranges of workers, resulting in API calls failing. Even the API gateway of Cloudflare AI has been blocked.
九Do you also work on AIGC-related products? Do you have any information on how this AI is charged? I couldn't find it in the documentation.
九i write an issue in github, hope can get answers.
https://github.com/cloudflare/cloudflare-docs/issues/11835
https://github.com/cloudflare/cloudflare-docs/issues/11835
GitHub
Which Cloudflare product(s) does this pertain to? Workers AI Subject Matter @cf/stabilityai/stable-diffusion-xl-base-1.0 Content Location Cloudflare’s AI is fantastic. I am testing the text-to-imag...
MI try to run a minecraft server through cloudflare tunnel.
How should the config be to allow all in/outbound on port 25566 to allow both TCP/UDP?
I have been clicking around, with rules, firewall DNS and egress policies but nothing works?
And how to disable the robot/captcha check to allow Minecraft to traffic with no throttle?
(Yes switched the default port to 25566)
How should the config be to allow all in/outbound on port 25566 to allow both TCP/UDP?
I have been clicking around, with rules, firewall DNS and egress policies but nothing works?
And how to disable the robot/captcha check to allow Minecraft to traffic with no throttle?
(Yes switched the default port to 25566)
H
D?tunnel-tcp
FCloudflare Tunnels use Cloudflare's proxy, which only supports proxying HTTP Traffic. If you want to use non-http applications over your tunnel, Cloudflare has a few other options:
For a few specific protocols such as SSH, RDP, and SMB, Cloudflare has guides for them here:
https://developers.cloudflare.com/cloudflare-one/applications/non-http/
For Arbitrary TCP like Minecraft, MySQL, and any other tcp application, Cloudflare has a guide here: https://developers.cloudflare.com/cloudflare-one/applications/non-http/arbitrary-tcp/
For Arbitrary UDP like Minecraft Bedrock, SMTP, and any other udp application, you will need to use Private Networking with WARP: https://developers.cloudflare.com/cloudflare-one/connections/connect-apps/private-net/connect-private-networks/
Please note for all of these except SSH and VNC which can be browser-rendered, you will either need to use cloudflared (Cloudflare's tunnel daemon) on the client machine running in the background or Private Networking with WARP, and have WARP installed on the client machine logged into your Zero Trust Team.
For a few specific protocols such as SSH, RDP, and SMB, Cloudflare has guides for them here:
https://developers.cloudflare.com/cloudflare-one/applications/non-http/
For Arbitrary TCP like Minecraft, MySQL, and any other tcp application, Cloudflare has a guide here: https://developers.cloudflare.com/cloudflare-one/applications/non-http/arbitrary-tcp/
For Arbitrary UDP like Minecraft Bedrock, SMTP, and any other udp application, you will need to use Private Networking with WARP: https://developers.cloudflare.com/cloudflare-one/connections/connect-apps/private-net/connect-private-networks/
Please note for all of these except SSH and VNC which can be browser-rendered, you will either need to use cloudflared (Cloudflare's tunnel daemon) on the client machine running in the background or Private Networking with WARP, and have WARP installed on the client machine logged into your Zero Trust Team.
DCould sound like you're maybe hitting the limitation, that is being mentioned in the last sentence?
MNo luck. Nmap says 20 ports are open on domain with no useable and the needed port are filtered
DRegarding the above, the sentence I'm referring to:
Have you installed cloudflared on the client machine (the one running the nmap), and connected that machine to the network?
Or otherwise connected the machine you run nmap on, to your Private Networking via your Zero Trust Team?
Please note for all of these except SSH and VNC which can be browser-rendered, you will either need to use cloudflared (Cloudflare's tunnel daemon) on the client machine running in the background or Private Networking with WARP, and have WARP installed on the client machine logged into your Zero Trust Team.
Have you installed cloudflared on the client machine (the one running the nmap), and connected that machine to the network?
Or otherwise connected the machine you run nmap on, to your Private Networking via your Zero Trust Team?
DCould maybe be a bit expensive to use at the moment though, considering the blog post 
:
That being said, - instead of starting the crowd sourcing, you might want to check: https://discord.com/channels/595317990191398933/812577823599755274/1093117323923173416
:At the moment of writing Free, Professional and Business accounts have access only to IP Lists.
That being said, - instead of starting the crowd sourcing, you might want to check: https://discord.com/channels/595317990191398933/812577823599755274/1093117323923173416
MNo the server is on cloudflared and i scan my doman
MRoot domain
MBut cloudflare declare my host name.damin / root domain
DThat is the problem, unless you point your domain to a local IP address that is available via your tunnel (and where every "client" is connected to your tunnel network), that isn't going to work.
MCreate a cname?
DIf you want to run a public server, that everyone on the Internet can join, and also have it behind Cloudflare, you would need something like e.g. Spectrum.
A CNAME, or any other DNS record(s), won't be able to help you with that.
A CNAME, or any other DNS record(s), won't be able to help you with that.
DYeah I tried that but it wasn't working..
DI added a second policy to allow that token, yes
DAllow, don't recall if it gave me any options to change that?
DAh, I see what you're saying