From a1558d119625b5edd01a7ef7bd35e27cc29d77da Mon Sep 17 00:00:00 2001 From: FAQ Bot Date: Sat, 15 Aug 2026 17:34:33 +0000 Subject: [PATCH 1/2] =?UTF-8?q?NEW:=20My=20Groq=20model=20stopped=20workin?= =?UTF-8?q?g=20=E2=80=94=20I'm=20getting=20a=20GroqError=20or=20model=5Fde?= =?UTF-8?q?commi?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- ...-decommissioned-replacement-and-reasonin.md | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) create mode 100644 _questions/llm-zoomcamp/module-1-rag/036_b391cb9b0b_groq-model-decommissioned-replacement-and-reasonin.md diff --git a/_questions/llm-zoomcamp/module-1-rag/036_b391cb9b0b_groq-model-decommissioned-replacement-and-reasonin.md b/_questions/llm-zoomcamp/module-1-rag/036_b391cb9b0b_groq-model-decommissioned-replacement-and-reasonin.md new file mode 100644 index 00000000..1bdd1946 --- /dev/null +++ b/_questions/llm-zoomcamp/module-1-rag/036_b391cb9b0b_groq-model-decommissioned-replacement-and-reasonin.md @@ -0,0 +1,18 @@ +--- +id: b391cb9b0b +question: My Groq model stopped working — I'm getting a GroqError or model_decommissioned + error. What do I do? +sort_order: 36 +--- + +As of August 2026, `llama-3.3-70b-versatile` and `qwen/qwen3-32b` are being retired by Groq. If you see errors like `GroqError` / `model_decommissioned`, switch to a currently supported model. + +Recommended replacement (as of the report): +- `MODEL = "openai/gpt-oss-120b"` + +Also check your request parameters: +- `gpt-oss-120b` accepts only `reasoning_effort` values: `"low"`, `"medium"`, or `"high"`. +- If your code uses `reasoning_effort="none"`, update it to `reasoning_effort="low"`. +- Otherwise you can get a validation error on every request. + +If you’re still failing after switching models, verify the exact model id you’re sending to Groq matches the new one you intend to use (including case/spelling) and that `reasoning_effort` is set to one of the supported values. \ No newline at end of file From 9c4adee43a6e3d509359134ba007d2fdfeae0597 Mon Sep 17 00:00:00 2001 From: Alexey Grigorev Date: Wed, 9 Sep 2026 14:36:39 +0200 Subject: [PATCH 2/2] Fix Groq decommissioned FAQ: Responses API snippet, links, question form (#356) --- ...decommissioned-replacement-and-reasonin.md | 31 +++++++++++++------ 1 file changed, 21 insertions(+), 10 deletions(-) diff --git a/_questions/llm-zoomcamp/module-1-rag/036_b391cb9b0b_groq-model-decommissioned-replacement-and-reasonin.md b/_questions/llm-zoomcamp/module-1-rag/036_b391cb9b0b_groq-model-decommissioned-replacement-and-reasonin.md index 1bdd1946..368f8055 100644 --- a/_questions/llm-zoomcamp/module-1-rag/036_b391cb9b0b_groq-model-decommissioned-replacement-and-reasonin.md +++ b/_questions/llm-zoomcamp/module-1-rag/036_b391cb9b0b_groq-model-decommissioned-replacement-and-reasonin.md @@ -1,18 +1,29 @@ --- id: b391cb9b0b -question: My Groq model stopped working — I'm getting a GroqError or model_decommissioned - error. What do I do? +question: How do I fix GroqError or model_decommissioned errors when my Groq model stops working? sort_order: 36 --- -As of August 2026, `llama-3.3-70b-versatile` and `qwen/qwen3-32b` are being retired by Groq. If you see errors like `GroqError` / `model_decommissioned`, switch to a currently supported model. +If you see `GroqError` / `model_decommissioned`, your Groq model ID was retired. As of August 2026, `llama-3.3-70b-versatile` and `qwen/qwen3-32b` were retired by Groq — switch to a currently supported model and update reasoning effort: -Recommended replacement (as of the report): -- `MODEL = "openai/gpt-oss-120b"` +```python +import os +from openai import OpenAI -Also check your request parameters: -- `gpt-oss-120b` accepts only `reasoning_effort` values: `"low"`, `"medium"`, or `"high"`. -- If your code uses `reasoning_effort="none"`, update it to `reasoning_effort="low"`. -- Otherwise you can get a validation error on every request. +client = OpenAI( + api_key=os.getenv("GROQ_API_KEY"), + base_url="https://api.groq.com/openai/v1", +) -If you’re still failing after switching models, verify the exact model id you’re sending to Groq matches the new one you intend to use (including case/spelling) and that `reasoning_effort` is set to one of the supported values. \ No newline at end of file +response = client.responses.create( + model="openai/gpt-oss-120b", + input="Hello", + reasoning={"effort": "low"}, # gpt-oss accepts only "low", "medium", or "high" +) + +print(response.output_text) +``` + +If your code uses `reasoning_effort="none"` (Chat Completions style, valid for older Qwen models), the Responses API equivalent is `reasoning={"effort": "low"}` — `"none"` is rejected for `gpt-oss-120b` and you get a validation error on every request. + +Check [Groq deprecations](https://console.groq.com/docs/deprecations) for current model status, [Groq reasoning docs](https://console.groq.com/docs/reasoning) for supported effort values, and [Groq Responses API docs](https://console.groq.com/docs/responses-api) for the request shape.