Hey @ajgeiss0702 . Which gpt model is that? And can you share your account ID with me (by DM) and I

Ssamjs Hey @ajgeiss0702 . Which gpt model is that? And can you share your account ID wi...

A

ajgeiss0702•9/17/25, 5:01 PM

its the 20b. my account id is f55b85c8a963663b11036975203c63c0

A

ajgeiss0702•9/17/25, 6:05 PM

ive been watching the ai gateway, and most errors are either capacity or something like

AiError: AiError: unexpected tokens remaining in message header: [\"……………...]???…….……...…..\", \"………...…ż…………..………………….………………………..\", \"...??…\",

AiError: AiError: unexpected tokens remaining in message header: [\"……………...]???…….……...…..\", \"………...…ż…………..………………….………………………..\", \"...??…\",

AiError: AiError: unexpected tokens remaining in message header: [\"……………...]???…….……...…..\", \"………...…ż…………..………………….………………………..\", \"...??…\",

AiError: AiError: unexpected tokens remaining in message header: [\"……………...]???…….……...…..\", \"………...…ż…………..………………….………………………..\", \"...??…\",

.... and it goes on for a while until it looks like some of the strings are words from thoughts?

A

ajgeiss0702•9/17/25, 6:06 PM

my retry logic automatically tries the 120b model if the 20b model fails after 8 retries, and that works first time every time ive seen

Aajgeiss0702 :/

.

.r20•9/18/25, 2:41 AM

yeah im running into the same issue as well im getting an absolute ton of 429 exceeded capacity errors when i try to use workersai

.

.r20•9/18/25, 2:44 AM

not sure if this is temporary since i only started using it today but its been like this all day

..r20 yeah im running into the same issue as well im getting an absolute ton of 429 ex...

A

ajgeiss0702•9/18/25, 2:45 AM

usually bge-m3 is fine for me, but I have seen a few errors with it today. probably just temporary

A

ajgeiss0702•9/18/25, 2:45 AM

capcity errors are pretty common on workers ai, especially with what i assume are the more popular models

.

.r20•9/18/25, 2:46 AM

ah i assume due to autoscaling serverless

V

Vikash•9/18/25, 5:32 AM

Getting too many errors: AiError: 3040: Capacity temporarily exceeded, please try again.

const embeddings: any = await env.AI.run("@cf/baai/bge-m3", {
    text: [prompt],
  });

const embeddings: any = await env.AI.run("@cf/baai/bge-m3", {
    text: [prompt],
  });

const embeddings: any = await env.AI.run("@cf/baai/bge-m3", {
    text: [prompt],
  });

const embeddings: any = await env.AI.run("@cf/baai/bge-m3", {
    text: [prompt],
  });

..r20 yeah im running into the same issue as well im getting an absolute ton of 429 ex...

V

Vikash•9/18/25, 5:42 AM

Hey, looks like you are using a gateway, how can I change my code to also use a gateway instead the direct reqeust?

const embeddings: any = await env.AI.run("@cf/baai/bge-m3", {
text: [text],
});

F

fortytwo•9/18/25, 6:15 AM

Can a folder within a bucket be used as a RAG source? Or do I need to create separate buckets for independent rag data?

VVikash Hey, looks like you are using a gateway, how can I change my code to also use a ...

.

.r20•9/18/25, 6:22 AM

https://developers.cloudflare.com/ai-gateway/usage/providers/workersai/

Cloudflare Docs

Workers AI

Use AI Gateway for analytics, caching, and security on requests to Workers AI. Workers AI integrates seamlessly with AI Gateway, allowing you to execute AI inference via API requests or through an environment binding for Workers scripts. The binding simplifies the process by routing requests through your AI Gateway with minimal setup.

L

Lami•9/18/25, 10:29 AM

@cf/google/embeddinggemma-300m

@cf/google/embeddinggemma-300m

@cf/google/embeddinggemma-300m

@cf/google/embeddinggemma-300m Does anyone know the price of this model?

VVikash Getting too many errors: AiError: 3040: Capacity temporarily exceeded, please t...

S

samjsOP•9/18/25, 3:26 PM

Hey @Vikash and others. Apologies for the 3040 errors you all are seeing. We're aware of these issues (we get alerted when they happen) and we're working to (a) provision more capacity for those models, and (b) implement some additional measures to keep capacity available even when we're getting flooded with requests. Appreciate your patience and thanks for flagging these issues!

Ffortytwo Can a folder within a bucket be used as a RAG source? Or do I need to create sep...

S

samjsOP•9/18/25, 6:15 PM

I believe we only support using the entire bucket. #autorag channel might know more about this however!

Ssamjs I believe we only support using the entire bucket. #autorag channel might know ...

F

fortytwo•9/18/25, 6:23 PM

Ah I missed that space. Thanks!

R

Razmjoo•9/18/25, 6:36 PM

Hi, is there any outages with AI Workers? I am getting a lot of errors on text-embeddings workers (using bge-m3) cc: @samjs

{
  "errors": [
    {
      "message": "AiError: AiError: model temporarily unavailable (58da3f75-5ff1-46ba-a45e-a59d4ada1c33)",
      "code": 9000
    }
  ],
  "success": false,
  "result": {},
  "messages": []
}

{
  "errors": [
    {
      "message": "AiError: AiError: model temporarily unavailable (58da3f75-5ff1-46ba-a45e-a59d4ada1c33)",
      "code": 9000
    }
  ],
  "success": false,
  "result": {},
  "messages": []
}

{
  "errors": [
    {
      "message": "AiError: AiError: model temporarily unavailable (58da3f75-5ff1-46ba-a45e-a59d4ada1c33)",
      "code": 9000
    }
  ],
  "success": false,
  "result": {},
  "messages": []
}

{
  "errors": [
    {
      "message": "AiError: AiError: model temporarily unavailable (58da3f75-5ff1-46ba-a45e-a59d4ada1c33)",
      "code": 9000
    }
  ],
  "success": false,
  "result": {},
  "messages": []
}

Ssamjs Hey @Vikash and others. Apologies for the 3040 errors you all are seeing. We're ...

I

Isaac McFadyen•9/18/25, 6:38 PM

@Razmjoo see this message

IIsaac McFadyen @Razmjoo see this message

R

Razmjoo•9/18/25, 6:41 PM

Are there any timeline e.g, do we expect it in coming hours to get fixed; or is it that might go over weekend/next week... just thinking if we need a fallback plan for business continuity ...

E

engnadeau•9/18/25, 7:08 PM

i thought i was going crazy on my end. is there not a status page for Workers AI?

E

engnadeau•9/18/25, 7:08 PM

AiError: 9000: model temporarily unavailable

AiError: 9000: model temporarily unavailable

AiError: 9000: model temporarily unavailable

AiError: 9000: model temporarily unavailable when trying to use

@cf/baai/bge-m3

@cf/baai/bge-m3

@cf/baai/bge-m3

@cf/baai/bge-m3

Eengnadeau `AiError: 9000: model temporarily unavailable` when trying to use `@cf/baai/bge-...

A

ajgeiss0702•9/18/25, 7:12 PM

im seeing this too

E

engnadeau•9/18/25, 7:24 PM

@samjs is the 9000 error related to the 3040 errors? is there a status page for Workers AI and model capacity? https://www.cloudflarestatus.com/ says all systems operational

Cloudflare Status

Welcome to Cloudflare's home for real-time and historical data on system performance.

S

samjsOP•9/18/25, 7:27 PM

The 9000 and 3040 errors are related -- we're seeing a huge influx of requests. Deploying some config changes now to try and help with that.

The follow up work I mentioned before should help us deal with this situations better in the future.

is there a status page for Workers AI and model capacity? https://www.cloudflarestatus.com/ says all systems operational

Currently the status page reflects the global status, not specific models.

J

James•9/18/25, 7:39 PM

Specific models being inoperable could result in 100% unavailability for some people's applications. I'd encourage y'all to reconsider your approach to status reporting here in future please so that these things can be better communicated to folks who might not just be in the Discord. :MeowHeartCloudflare:

S

samjsOP•9/18/25, 7:53 PM

https://www.cloudflarestatus.com/incidents/nrxwc9tny0z7

Workers AI Errors for bge-m3 Model

�

𝗞enneth•9/18/25, 8:16 PM

Hi Cloudflare team,

I've been working with AutoRAG and it's a great feature for building RAG systems. However, I notice it's currently limited to models available in Workers AI.
Is there any plan to support custom model providers and custom embedding providers in AutoRAG?

Specifically:
Custom LLM providers (OpenAI, Claude, ETC)
Custom embedding providers (OpenAI embeddings, specialized models)
Bring-your-own-model support

This would enable using specialized embeddings for specific domains, cost optimization across providers, and better architectural flexibility for enterprise use cases.
Is this being considered for the roadmap, or is there currently an undocumented way to achieve this?

Thanks for the great work on Workers AI.

�𝗞enneth Hi Cloudflare team, I've been working with AutoRAG and it's a great feature fo...

C

Chaika•9/18/25, 8:39 PM

I would ask in #autorag , it's a bit weird to ask in the workers-ai channel about not using workers-ai lol

CChaika I would ask in #autorag , it's a bit weird to ask in the workers-ai channel abou...

S

samjsOP•9/18/25, 9:00 PM

Haha, that's okay we have thick skin

Does sound like a reasonable use case, but indeed you'll get a better response in the #autorag channel!

�

𝗞enneth•9/18/25, 9:20 PM

@samjs Sorry for posting in the wrong channel! I meant to ask this in #autorag. Thanks for pointing that out.

A

achesui•9/18/25, 11:38 PM

Hello, I'm not being able to make the new model

nova-3

nova-3

nova-3

nova-3 work with language spanish websocket via REST API, i'm testing this with the repository: https://github.com/cloudflare/realtime-examples/tree/main/ai-tts-stt when setting the language to 'es' or 'es-419' it gives me a 500 error: Failed to establish Nova WebSocket: 500 while the websocket keeps running, it works for english only, can anyone help me?

GitHub

realtime-examples/ai-tts-stt at main · cloudflare/realtime-examples

Contribute to cloudflare/realtime-examples development by creating an account on GitHub.

Ffortytwo Ah I missed that space. Thanks!

.

.r20•9/19/25, 7:56 PM

yeah itll just index the entire bucket not just the folder

..r20 yeah itll just index the entire bucket not just the folder

F

fortytwo•9/19/25, 8:11 PM

I did notice you can filter on directories. But not sure when multiple rag source buckets vs one bucket + multiple dirs + dir filters would be best

.

.r20•9/19/25, 8:40 PM

idt it matters id just do the directory that’s easiest

.

.r20•9/19/25, 8:40 PM

but i think autorag has some fundamental issues they need to fix so i would rather use vectorize directly its better

F

fortytwo•9/19/25, 8:46 PM

What issues have you seen there?

E

engnadeau•9/19/25, 8:51 PM

error of the day:

  error: '4002: could not route request to AI model',

error: '4002: could not route request to AI model',

  error: '4002: could not route request to AI model',

error: '4002: could not route request to AI model', with

  model: '@cf/baai/bge-m3',

model: '@cf/baai/bge-m3',

  model: '@cf/baai/bge-m3',

model: '@cf/baai/bge-m3',

is this also due to high demand?

E

engnadeau•9/19/25, 9:07 PM

is Workers AI considered production ready?

J

James•9/19/25, 9:42 PM

https://www.cloudflarestatus.com/incidents/k9wl5k2hb13m

Workers AI experiencing degraded availability in some models

Ffortytwo What issues have you seen there?

.

.r20•9/19/25, 9:48 PM

workers ai is still a few iterations away from production ready i cant launch it to my customers in the state its in rn (although they are working on it p fast), and indexing isnt super reliable rn imo (partially due to workers ai and then other generic errors it throws)

.

.r20•9/19/25, 9:49 PM

for prototyping its fine but vectorize is very good and i still recommend that and not a lot more work than autorag

F

fortytwo•9/19/25, 9:49 PM

Yeah I did have a bunch of indexing errors when I pushed stuff, had to redo it 5 or 6 times to get all of them done

.

.r20•9/19/25, 9:49 PM

yeah the problem is my customers update the rag extremely frequently so i cant rely on their indexing, but like i said its still good for prototyping

F

fortytwo•9/19/25, 9:55 PM

Ok, well that's good to know. So you're using vectorize "manually", basically? Still storing data in a cloudflare bucket but generating and searching the vectors yourself rather than the autorag apis?

.

.r20•9/19/25, 10:38 PM

i store data in d1 and yeah i just make the embeddings through openai and insert into the vectorize

.

.r20•9/19/25, 10:38 PM

and then searching isnt that hard either

.

.r20•9/19/25, 10:39 PM

u can vibe code most of this stuff anyway through cloudflare's mcp server and whatever ai powered ide u use

R

ray•9/21/25, 4:23 AM

Is any one else getting issues with tool calls? I can only seem to get proper tool call responses via the ai sdk when I use mistral-small-3.1-24b-instruct

R

ray•9/21/25, 4:24 AM

All the other tool calling models either return a text type with tool params json stringified or error out

A

Ashkan•9/21/25, 9:53 AM

Hello, when I want to poll batch request I encounter this error:

Error: 3030: 1 validation error for VllmBatchRequest

Error: 3030: 1 validation error for VllmBatchRequest

Error: 3030: 1 validation error for VllmBatchRequest

Error: 3030: 1 validation error for VllmBatchRequest does anyone know what doesn't mean and how to resolve it?

Hey @ajgeiss0702 . Which gpt model is that? And can you share your account ID with me (by DM) and I

Similar Threads