From e6fb2a51e264378e042e6e00c202a4b2f98e8d09 Mon Sep 17 00:00:00 2001
From: Bola Malek <bola@baseten.co>
Date: Fri, 1 Aug 2025 16:09:59 -0700
Subject: [PATCH] Add Qwen3 with BIS

---
 qwen/qwen-3-30B-A3/config.yaml | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)
 create mode 100644 qwen/qwen-3-30B-A3/config.yaml

diff --git a/qwen/qwen-3-30B-A3/config.yaml b/qwen/qwen-3-30B-A3/config.yaml
new file mode 100644
index 000000000..dea620649
--- /dev/null
+++ b/qwen/qwen-3-30B-A3/config.yaml
@@ -0,0 +1,22 @@
+model_name: qwen-30b-BIS
+python_version: py39
+resources:
+  accelerator: H100
+  cpu: '1'
+  memory: 10Gi
+  use_gpu: true
+trt_llm:
+  build:
+    base_model: decoder
+    checkpoint_repository:
+      repo: Qwen/Qwen3-30B-A3B-Instruct-2507
+      revision: main
+      source: HF
+  inference_stack: v2
+  runtime:
+    enable_chunked_prefill: true
+    max_batch_size: 128
+    max_num_tokens: 8192
+    max_seq_len: 131072
+    served_model_name: Qwen/Qwen3-30B-A3B-Instruct-2507
+    tensor_parallel_size: 1