Foundry Tools
Formerly known as Azure AI Services or Azure Cognitive Services is a unified collection of prebuilt AI capabilities within the Microsoft Foundry platform
Deployed gpt-5.2 to foundry running into rate limit issues or very slow responses
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(268.8, 22%, 36%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EG%3C/text%3E%3C/svg%3E)
GS
430
Reputation points
Hello,
I deployed gpt-5.2 to azure -
My rate limit is set pretty high - I am the only user testing this right now and am getting
{"type":"response.failed","sequence_number":3,"response":{"id":"resp_077206c4b441379001693c2b62d0308196afcbb0353a0c03ce","object":"response","created_at":1765550946,"status":"failed","background":false,"content_filters":null,"error":{"code":"rate_limit_exceeded","message":" | ==================== d001-20251211012732-api-default-78bd44c5dc-9w645 ====================\n | Traceback (most recent call last):\n | \n | File "/usr/local/lib/python3.12/site-packages/inference_server/routes.py", line 726, in streaming_completion\n | await response.write_to(reactor)\n | \n | oai_grpc.errors.ServerError: | no_kv_space\n | "},
Last night the same deployment was working. Today the responses are either slow or I get a rate limit exceed on a simple prompt like Hello
Is it too early to use gpt-5.2 in Microsoft Foundry?
Foundry Tools
-
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(12.8, 19%, 11%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EHS%3C/text%3E%3C/svg%3E)
-
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(217.60000000000002, 19%, 31%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ETA%3C/text%3E%3C/svg%3E)
Thiago Almeida 25 Reputation points2025-12-12T21:15:30.9366667+00:00 Same here. Same history. When I first deployed the model, it was working. Now it's not. I guess it is a bit early for gpt-5.2 on azure at the moment.
-
GS 430 Reputation points
2025-12-13T04:26:41.15+00:00 @Anonymous things seemed to be working for a little bit and now they are back to not working
-
John Vance 5 Reputation points2025-12-14T03:11:43.9+00:00 I am seeing the same thing this evening after everything was working fine yesterday. No config changes in Azure or the client. I suspect a back-end Azure issue.
-
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(243.2, 90%, 33%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ES%3C/text%3E%3C/svg%3E)
Sellers 15 Reputation points2025-12-14T15:10:15.3233333+00:00 It is absolutely unacceptable this is still happening.
-
GS 430 Reputation points
2025-12-15T02:48:55.85+00:00 yea just got a 429 capacity error again and a message about capacity not being there .. what a shame .. I am surprised that not more people are talking about this.
-
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(57.599999999999994, 36%, 15%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ECL%3C/text%3E%3C/svg%3E)
Cynthia Luijkx 50 Reputation points2025-12-15T07:59:56.3366667+00:00 Experiencing the same here, even after the weekend :(
-
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(316.8, 84%, 40%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EDW%3C/text%3E%3C/svg%3E)
David West 5 Reputation points2025-12-15T13:52:52.9966667+00:00 Running into the same issue here. It seemed like it was working over the weekend some, but it's back to failure this morning.
-
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(150.4, 28.999999999999996%, 24%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EAL%3C/text%3E%3C/svg%3E)
Adam Lass 0 Reputation points2025-12-16T11:43:55.2333333+00:00 Same issue here.
Question:
Are you guys using Semantic Kernel or Microsoft Agent Framework?We are on SK currently, but want to upgrade at some point. Don't know if this can introduce some incompatibility in the future, like the issues we are seeing right now?
-
Cynthia Luijkx 50 Reputation points
2025-12-16T12:02:38.1866667+00:00 @Adam Lass We are on Microsoft Agent Framework. I've also seen these issues when I call an API directly without any of the frameworks around it so doesn't really seem related to that.
In general I am very happy with our migration from Semantic Kernel to Agent Framework though, would recommend that :)
-
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(195.2, 94%, 28%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EHJ%3C/text%3E%3C/svg%3E)
-
Cynthia Luijkx 50 Reputation points
2025-12-18T14:51:34.3133333+00:00 Also seems like the prompt caching is not working properly; at times where I do get a response I don't see any cached tokens in the usage. I can imagine that doesn't help the latency either.
-
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(217.60000000000002, 19%, 31%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EMS%3C/text%3E%3C/svg%3E)
Michael Streif 25 Reputation points2026-01-02T09:55:17.4166667+00:00 We are also experiencing the same issue as the other comments. We already increased the quota of our GPT 5.2 deployment to 10 million TPM, but we still randomly get the 429 error response. I am currently using it in our development environment and already there we have these problems which makes it unusable for our production environment and real usage with our customers. This is a big problem!
-
GS 430 Reputation points
2026-01-02T15:21:13.6266667+00:00 @Michael Streif yes ..i have not run into the 429 again, but it is super slow - we deployed it to our enterprise, with a disclaimer that this is not quite usable yet ( In Azure ). 5.1 is still our model of choice for now
Sign in to comment
Sign in to answer