I don't think there's any public models

T

ToriannaOP•3/14/25, 10:58 AM

I know Google is working on one, but that's internal

K

karenr•3/14/25, 11:01 AM

Google - oh really?
yeah.. current models are multimodal AI model with both vision and language capabilities not specific to compression

TTorianna I don't think there's any public models

I

Isaac McFadyen•3/14/25, 3:24 PM

Nvidia has one and it's teeeechnically public but not really? i.e. it's integrated into their new Neural Shader stuff but I don't believe the model is open-weight.

IIsaac McFadyen CC @michelle ^ not 100% sure that's what's happening, but since multiple users a...

D

dmaretskyi•3/14/25, 3:25 PM

do you have a public bug tracker for this?

D

dmaretskyi•3/14/25, 3:25 PM

just so I can check when this gets resolved

Ddmaretskyi do you have a public bug tracker for this?

I

Isaac McFadyen•3/14/25, 3:25 PM

Nope but if I don't hear back from Michelle in a while I can try to escalate to others.

D

dmaretskyi•3/14/25, 3:25 PM

okay, thanks for your help!

Ddmaretskyi okay, thanks for your help!

I

Isaac McFadyen•3/14/25, 3:26 PM

I'm not sure if R1 works well with tools at all, based on what I've heard. But the missing opening <think> is probably a bug, I can't imagine it's intended.

J

JoaoCupido•3/14/25, 7:45 PM

Flux Schnell model still doesn't have a random seed implementation and image sizes (width x height), unlike Stable Diffusion models. When is this going to be implemented, if possible?

B

blindChicken•3/16/25, 5:58 PM

Has there been a change to the

hermes-2-pro-mistral-7b

hermes-2-pro-mistral-7b

model. I have a very basic app based on CF's function calling tutorial. I made the app about a month ago and has been working fine since. But over the past few days it has stopped working and it appears that the model has changed the format of its response. Here is how it is responding now:

User Prompt: show me the color of an avocado?
AI result is: {
  response: '<tool_call>\n' +
    "{'arguments': {'r': 69, 'g': 139, 'b': 69}, 'name': 'switchLightColor'}\n" +
    '</tool_call>\n' +
    '\n' +
    'I have chosen a color that represents an avocado. The color of an avocado is typically a pale green with a hint of creaminess. To represent this, I have selected a color with a higher green value (139) and balanced amounts of red (69) and blue (69) values. This creates a color that is reminiscent of the natural hue of an avocado, bringing the essence of the fruit into the room.',
  usage: { prompt_tokens: 0, completion_tokens: 0, total_tokens: 0 }
}

User Prompt: show me the color of an avocado?
AI result is: {
  response: '<tool_call>\n' +
    "{'arguments': {'r': 69, 'g': 139, 'b': 69}, 'name': 'switchLightColor'}\n" +
    '</tool_call>\n' +
    '\n' +
    'I have chosen a color that represents an avocado. The color of an avocado is typically a pale green with a hint of creaminess. To represent this, I have selected a color with a higher green value (139) and balanced amounts of red (69) and blue (69) values. This creates a color that is reminiscent of the natural hue of an avocado, bringing the essence of the fruit into the room.',
  usage: { prompt_tokens: 0, completion_tokens: 0, total_tokens: 0 }
}

User Prompt: show me the color of an avocado?
AI result is: {
  response: '<tool_call>\n' +
    "{'arguments': {'r': 69, 'g': 139, 'b': 69}, 'name': 'switchLightColor'}\n" +
    '</tool_call>\n' +
    '\n' +
    'I have chosen a color that represents an avocado. The color of an avocado is typically a pale green with a hint of creaminess. To represent this, I have selected a color with a higher green value (139) and balanced amounts of red (69) and blue (69) values. This creates a color that is reminiscent of the natural hue of an avocado, bringing the essence of the fruit into the room.',
  usage: { prompt_tokens: 0, completion_tokens: 0, total_tokens: 0 }
}

User Prompt: show me the color of an avocado?
AI result is: {
  response: '<tool_call>\n' +
    "{'arguments': {'r': 69, 'g': 139, 'b': 69}, 'name': 'switchLightColor'}\n" +
    '</tool_call>\n' +
    '\n' +
    'I have chosen a color that represents an avocado. The color of an avocado is typically a pale green with a hint of creaminess. To represent this, I have selected a color with a higher green value (139) and balanced amounts of red (69) and blue (69) values. This creates a color that is reminiscent of the natural hue of an avocado, bringing the essence of the fruit into the room.',
  usage: { prompt_tokens: 0, completion_tokens: 0, total_tokens: 0 }
}

The tool call is being returned within the response and is correctly tagged with

<tool_call>

<tool_call>

but the response no longer returns a

tool_calls

tool_calls

prop, so my code isn't calling anything it only shows the malformed response.

B

blindChicken•3/16/25, 6:11 PM

If I replace

hermes-2-pro-mistral-7b

hermes-2-pro-mistral-7b

with

@cf/meta/llama-3.3-70b-instruct-fp8-fast

@cf/meta/llama-3.3-70b-instruct-fp8-fast

then it works fine. Something must have changed with the Mistral model

K

kalepail•3/16/25, 7:58 PM

I'm trying to use the

@cf/meta/llama-guard-3-8b

@cf/meta/llama-guard-3-8b

. What's the proper way to tweak this to modify the

UNSAFE CONTENT CATEGORIES

UNSAFE CONTENT CATEGORIES

? A lot of sexually explict stuff is getting through when the prompts aren't very long. I want a really robust NSFW content filter but asis this model isn't working well

BblindChicken Has there been a change to the `hermes-2-pro-mistral-7b` model. I have a very ba...

J

James•3/16/25, 8:26 PM

cc @michelle possible breaking change or incident with tool calling for this model?

J

James•3/16/25, 8:44 PM

@Isaac | AS213339 confirmed the regression and has raised an escalation with the team. Thanks for the report.

JJames @Isaac | AS213339 confirmed the regression and has raised an escalation with the...

I

Isaac McFadyen•3/16/25, 8:45 PM

^ Oh yes, thanks for mentioning; forgot to say that here.

H

Houston CF•3/16/25, 9:58 PM

@blindChicken @Isaac | AS213339 @James Are we seeing the results we expected now? We have deployed a fix for the " tool calling using hermes-2-pro-mistral-7b model" issue.

J

James•3/16/25, 9:59 PM

Looks good now, thank you!

JJames Looks good now, thank you!

B

blindChicken•3/17/25, 1:17 AM

We just tested it on our end and it appears to be working correctly. Thanks!

M

markvaldez•3/17/25, 6:11 PM

Hi, I don't see any way to rotate my REST bearer AI token..

M

markvaldez•3/17/25, 6:14 PM

I didn't create a token, it was created for me on the dashboard when I went to the AI Worker page

M

markvaldez•3/17/25, 6:39 PM

I applied as an Engineer at Cloudflare this past month, but I don't think I got past the resume screen. I live around the corner from the SF offices and have been there for a hackathon and lunch-and-learn HONO event the past 2 weeks. I can fix this issue: https://github.com/valdezm/dash.cloudflare.com/issues/1

GitHub

Cannot rotate REST AI Bearer token · Issue #1 · valdezm/dash.cloudf...

Problem A REST Bearer token was created for me and does not show up in my Account API Tokens using these instructions: https://developers.cloudflare.com/workers-ai/get-started/rest-api/#1-get-api-t...

M

markvaldez•3/17/25, 6:40 PM

https://tenor.com/view/bills-ugh-gif-9811803164280522639

Tenor

M

markvaldez•3/17/25, 6:41 PM

Send me a DM and I will happliy get to work

M

markvaldez•3/17/25, 6:43 PM

Also, you all are awesome

A

Alex Patterson•3/20/25, 3:30 AM

Seems like all the models on Workers AI have such small Context WIndows, is there any that will be added that come closer to Google Gemini?

S

si(dot).xth•3/20/25, 11:30 AM

my pillow fixed my issue:

raw data for simplicity and efficiency in handling large images, this method may lead to more API calls if the image is re-generated each time it's accessed (Caching handles that last part.)

base64 encoding to minimize API calls over performance (Easier to implement)

Raw data + cache is the only way to go

Original: Blob not supported for workers ai? URL.CreateObjectURL() is not implemented

AAlex Patterson Seems like all the models on Workers AI have such small Context WIndows, is ther...

S

Silvan•3/20/25, 12:01 PM

Having the same issue here. Trying to create a story is fairly hard this way but some models do have bigger context windows but many of the "good ones" are less than 10k

J

JMPA•3/20/25, 9:14 PM

hello, is inpaiting ai workers working currently?

J

JMPA•3/20/25, 9:15 PM

i tried 2 models and they dont works.

J

JMPA•3/20/25, 9:16 PM

stable-diffusion-v1-5-inpainting
API response status: 400 body: {"errors":[{"message":"AiError: Bad input: Error: Type mismatch of '/mask', 'array' not in 'string'","code":5006}],"success":false,"result":{},"messages":[]}
API error: 400 - {"errors":[{"message":"AiError: Bad input: Error: Type mismatch of '/mask', 'array' not in 'string'","code":5006}],"success":false,"result":{},"messages":[]}
Response with status 400 in 1363 ms.

J

JMPA•3/20/25, 10:19 PM

anyone there?

J

James•3/20/25, 10:27 PM

Is this something that worked previously and now doesn't? Can you provide any code to reproduce?

JJMPA hello, is inpaiting ai workers working currently?

R

Raylight•3/22/25, 6:59 AM

As James implied, it would be easier to help if you have a snippet showing the call to the model, but in general: The docs show the same set of parameters (e.g.

image

image

,

mask

mask

) for several models, but the only model that does inpainting is

@cf/runwayml/stable-diffusion-v1-5-inpainting

@cf/runwayml/stable-diffusion-v1-5-inpainting

.
Make sure to supply both

image

image

and

mask

mask

when calling it. The format is

mask: [...new Uint8Array(await maskImage.arrayBuffer())]

mask: [...new Uint8Array(await maskImage.arrayBuffer())]

, where

maskImage

maskImage

could be the result of e.g.

await fetch(/* mask image url */)

await fetch(/* mask image url */)

.

S

Silvan•3/23/25, 11:09 AM

I really hope we get some “better” text generation models on here soon. i tried every single one for story creation and nothing came even remotely close to claude atm.

O

ochrons•3/23/25, 3:37 PM

I'm using

@cf/openai/whisper-large-v3-turbo

@cf/openai/whisper-large-v3-turbo

to convert audio into text. It works fine when recording on Pixel 9 in AAC format, but fails when recording on an older Oneplus 7T with AAC (I'm using Flutter to build the app). Using an AMR WB format works fine, though. Also all the AAC formats work fine when given to Google Gemini Flash/Pro models

O

ochrons•3/23/25, 3:38 PM

I tried both

MPEG-4 AAC Low complexity

MPEG-4 AAC Low complexity

and

MPEG-4 High Efficiency AAC

MPEG-4 High Efficiency AAC

formats with identical results

O

ochrons•3/23/25, 3:54 PM

ffprobe

ffprobe

says this about the one recorded on Oneplus,

[aac @ 0x560e66ce38c0] Number of scalefactor bands in group (46) exceeds limit (43).

[aac @ 0x560e66ce38c0] Number of scalefactor bands in group (46) exceeds limit (43).

while there are no errors with the AAC coming from pixel 9

0

0xCC•3/25/25, 4:17 AM

Hi the Daily Usage Limits for ai workder is not reset today
(my account 4631e4e4e57a2e43b16a0a5a1490d7c3 )

Original message was deleted

D

DaniFoldi•3/25/25, 8:35 AM

Argh I thought I got them all, thanks

N

nuno•3/25/25, 10:40 AM

Hey guys, I'm trying to generate embeddings using the

@cf/baai/bge-m3

@cf/baai/bge-m3

model but I keep getting the error below. Previously I was using the

@cf/baai/bge-base-en-v1.5

@cf/baai/bge-base-en-v1.5

model and everything was just fine.

err : AiError: 5006: Error: required properties at '/' are 'contexts'

err : AiError: 5006: Error: required properties at '/' are 'contexts'

err : AiError: 5006: Error: required properties at '/' are 'contexts'

err : AiError: 5006: Error: required properties at '/' are 'contexts'

Does anyone have any idea why is this happening? Any guidance here is much appreciated.

/

/dev/null•3/27/25, 9:51 AM

the cost of flux schnell is 172 neurons per image when I set with 1 step.

C

cyrilis•3/27/25, 2:22 PM

the deepseek model:

@cf/deepseek-ai/deepseek-r1-distill-qwen-32b

@cf/deepseek-ai/deepseek-r1-distill-qwen-32b

response sometime got empty response:

data: {"response":""}

data: {"response":""}

data: {"response":"","usage":{"prompt_tokens":0,"completion_tokens":1,"total_tokens":1}}

data: {"response":""}

data: {"response":""}

data: {"response":"","usage":{"prompt_tokens":0,"completion_tokens":1,"total_tokens":1}}

data: {"response":""}

data: {"response":""}

data: {"response":"","usage":{"prompt_tokens":0,"completion_tokens":1,"total_tokens":1}}

data: {"response":""}

data: {"response":""}

data: {"response":"","usage":{"prompt_tokens":0,"completion_tokens":1,"total_tokens":1}}

and sometimes it works all right, but missing the opening <think> tag. any others rushed into same problem?

Ccyrilis the deepseek model: `@cf/deepseek-ai/deepseek-r1-distill-qwen-32b` response s...

C

cyrilis•3/27/25, 2:24 PM

Or maybe it's just because I enabled smart placement? and a few AI models are not ready in some region ?

Ccyrilis the deepseek model: `@cf/deepseek-ai/deepseek-r1-distill-qwen-32b` response s...

I

Isaac McFadyen•3/27/25, 3:21 PM

missing the opening <think> tag

This is due to a quirk in the prompt template, which forces the <think> tag in order to require the model to think through it's response, but which results in this behavior

I

Isaac McFadyen•3/27/25, 3:21 PM

@michelle ^ been reported many times here, might want to fix it eventually

I

Isaac McFadyen•3/27/25, 3:22 PM

Not sure if it was always the case or whether it was a change but regardless, it's different from how other providers generally do it

P

Pytato•3/28/25, 6:58 AM

I think I have found a bug with Cloudflare's Worker Ai

I'm basically making a super simple Stable Diffusion wrapper with Pages and Worker Ai. While making the UI and testing it, I probably made 50 - 80
50 - 80
50 - 80
50 - 80 images with it. But my Daily Neurons Limits is still 0
0
0
0 (as in nothing used)

So, I checked the Pricing and found no mention of Stable Diffusion's neuron pricing.

Could it be they forgot to consider these model to add to the daily usage limit?
OR is 80
80
80
80 images too low to be show in the dashboard?

P

Pytato•3/28/25, 7:03 AM

Oh... I'm stupid

M

mr.niko.la•3/28/25, 2:56 PM

@cf/openai/whisper-large-v3-turbo

@cf/openai/whisper-large-v3-turbo

is there no sample worker code for this for something basic like r2 audio file to transcribe ?

I am using a worker to send it to 3rd party api to transcribe using queue

But I would like use worker ai whisper turbo as a failover. Without over engineering or breaking my webapp.

Do I just implement a new queue instead of a separate worker?

Perhaps eventually cf workflow but for now I don’t want to break my current webapp by reengineering anything too much.

Just want to do backup for the rare times 3rd api fails.

I don't think there's any public models

Similar Threads

Similar Threads

Similar Threads