[None][doc] Add deploy guide for Minimax M3

docs/source/_static/config_db.json

@@ -167,6 +167,18 @@
       "model_display_name": "Llama 4 Scout (FP8)",
       "model_url": "https://huggingface.co/nvidia/Llama-4-Scout-17B-16E-Instruct-FP8",
       "scenario": "Max Throughput"
+    },
+    {
+      "command": "trtllm-serve MiniMaxAI/MiniMax-M3 --config ${TRTLLM_DIR}/examples/configs/curated/minimax-m3-throughput.yaml",
+      "config_filename": "minimax-m3-throughput.yaml",
+      "config_github_url": "https://github.com/NVIDIA/TensorRT-LLM/blob/main/examples/configs/curated/minimax-m3-throughput.yaml",
+      "config_path": "examples/configs/curated/minimax-m3-throughput.yaml",
+      "config_raw_url": "https://raw.githubusercontent.com/NVIDIA/TensorRT-LLM/main/examples/configs/curated/minimax-m3-throughput.yaml",
+      "gpu_compatibility": "GB200",
+      "model": "MiniMaxAI/MiniMax-M3",
+      "model_display_name": "MiniMax-M3 (BF16)",
+      "model_url": "https://huggingface.co/MiniMaxAI/MiniMax-M3",
+      "scenario": "Max Throughput"
     }
   ],
   "entries": [
@@ -3504,6 +3516,10 @@
     }
   ],
   "models": {
+    "MiniMaxAI/MiniMax-M3": {
+      "display_name": "MiniMax-M3 (BF16)",
+      "url": "https://huggingface.co/MiniMaxAI/MiniMax-M3"
+    },
     "Qwen/Qwen3-30B-A3B": {
       "display_name": "Qwen3-30B-A3B",
       "url": "https://huggingface.co/Qwen/Qwen3-30B-A3B"

docs/source/deployment-guide/index.rst

@@ -37,3 +37,4 @@ The deployment guides below provide more detailed instructions for serving speci
    deployment-guide-for-qwen3.5-on-trtllm.md
    deployment-guide-for-kimi-k2-thinking-on-trtllm.md
    deployment-guide-for-glm-5-on-trtllm.md
+   deployment-guide-for-minimax-m3-on-trtllm.md

docs/source/models/supported-models.md

@@ -84,7 +84,7 @@ Note: Support for other models may vary. Features marked "N/A" are not applicabl
 [^8]: Supports text and image inputs. The vision tower runs in BF16 even when the text decoder is quantized (FP8 block-scale or NVFP4). The text decoder is also usable standalone (text-only) via the `Step3p5ForCausalLM` architecture.
 [^9]: Audio modality only supported on E2B/E4B variants.
 [^10]: Audio requires a checkpoint with a `sound_config` and is supported only on the full (non-disaggregated) model path, not the EPD disaggregated path.
-[^11]: Supports text, image, and video inputs over the block-sparse attention path. The published MXFP8 checkpoint is dequantized on load so the runtime sees an effectively BF16 model. The text decoder is also usable standalone (text-only) via the `MiniMaxM3SparseForCausalLM` architecture. KV cache reuse and MTP are not supported on the sparse-attention path in this release.
+[^11]: Supports text, image, and video inputs over the block-sparse attention path. The model is served in BF16. The text decoder is also usable standalone (text-only) via the `MiniMaxM3SparseForCausalLM` architecture. KV cache reuse and MTP are not supported on the sparse-attention path in this release.
 
 # Multimodal Feature Support Matrix (PyTorch Backend)
 

examples/configs/curated/lookup.yaml

@@ -81,3 +81,8 @@
   scenario: Max Throughput
   gpu_compatibility: "B200, GB200"
   disagg: true
+- model: MiniMaxAI/MiniMax-M3
+  arch: MiniMaxM3SparseForCausalLM
+  config_path: examples/configs/curated/minimax-m3-throughput.yaml
+  scenario: Max Throughput
+  gpu_compatibility: "GB200"

examples/configs/curated/minimax-m3-throughput.yaml

@@ -0,0 +1,15 @@
+max_batch_size: 256
+max_num_tokens: 8192
+tensor_parallel_size: 8
+moe_expert_parallel_size: 8
+enable_attention_dp: true
+trust_remote_code: true
+cuda_graph_config:
+  enable_padding: true
+kv_cache_config:
+  free_gpu_memory_fraction: 0.3
+  enable_block_reuse: false
+sparse_attention_config:
+  algorithm: minimax_m3
+stream_interval: 10
+num_postprocess_workers: 4

scripts/generate_config_table.py

@@ -90,6 +90,10 @@
         "display_name": "Kimi-K2-Thinking (NVFP4)",
         "url": "https://huggingface.co/nvidia/Kimi-K2-Thinking-NVFP4",
     },
+    "MiniMaxAI/MiniMax-M3": {
+        "display_name": "MiniMax-M3 (BF16)",
+        "url": "https://huggingface.co/MiniMaxAI/MiniMax-M3",
+    },
 }