From 888d12d3e194f82adff0d23f2254934b28b9281a Mon Sep 17 00:00:00 2001 From: Brian Nguyen Date: Fri, 30 Oct 2026 13:13:40 -0500 Subject: [PATCH 1/2] [None][doc] Point to day-0 side branches from the supported models page Muse Spark Open, MiniMax-M3.1-Flash, and Beam (Reflection AI) are currently supported on dedicated feature branches ahead of their merge into main. Add a temporary section to the supported-models page linking each side branch, its deployment guide, and the release container published from it, so users can find the support before it lands in a regular release. Entries are removed once a model's support is fully merged into main. Signed-off-by: Brian Nguyen --- docs/source/models/supported-models.md | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+) diff --git a/docs/source/models/supported-models.md b/docs/source/models/supported-models.md index 1c01d115f457..1afacf769864 100644 --- a/docs/source/models/supported-models.md +++ b/docs/source/models/supported-models.md @@ -57,6 +57,26 @@ The following is a table of supported models for the PyTorch backend: | `WhisperForConditionalGeneration` | Whisper | `openai/whisper-large-v3` | +## Models Supported on Day-0 Side Branches + +Some newly released models are first supported on a dedicated feature branch, with release +containers published from that branch, before the support is merged into `main`. The entries +below are temporary: each entry is removed once the model's support, including its deployment +guide, is fully merged into `main` and available in a regular release. Until then, use the +linked branch and container rather than `main`. + +The linked containers are preview releases published from the side branch and are retained on +NGC for roughly two months. If a tag is no longer available, check the +[release tags page](https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tensorrt-llm/containers/release/tags) +for a newer tag published from the same branch, or build the container from the branch sources +following its [Docker guide](https://github.com/NVIDIA/TensorRT-LLM/blob/main/docker/README.md). + +| Model | Side branch | Deployment guide | Release container | +| ----- | ----------- | ---------------- | ----------------- | +| Muse Spark Open | [`feat/musespark-open`](https://github.com/NVIDIA/TensorRT-LLM/tree/feat/musespark-open) | [Muse Spark Open deployment guide](https://github.com/NVIDIA/TensorRT-LLM/blob/feat/musespark-open/docs/source/deployment-guide/deployment-guide-for-musespark-open-on-trtllm.md), [speculative decoding guide](https://github.com/NVIDIA/TensorRT-LLM/blob/feat/musespark-open/docs/source/deployment-guide/deployment-guide-for-musespark-open-speculative-decoding-on-trtllm.md) | `nvcr.io/nvidia/tensorrt-llm/release:1.4.0a0.dev202610171236` | +| MiniMax-M3.1-Flash | [`feat/minimax-m3.1-flash`](https://github.com/NVIDIA/TensorRT-LLM/tree/feat/minimax-m3.1-flash) | [Serving MiniMax-M3.1-Flash](https://github.com/NVIDIA/TensorRT-LLM/blob/feat/minimax-m3.1-flash/docs/source/deployment-guide/deployment-guide-for-minimax-m3-on-trtllm.md#serving-minimax-m31-flash) in the MiniMax-M3 deployment guide | `nvcr.io/nvidia/tensorrt-llm/release:1.4.0a0.dev202610230958` | +| Beam (Reflection AI) | [`feat/reflection-beam`](https://github.com/NVIDIA/TensorRT-LLM/tree/feat/reflection-beam) | [Beam deployment guide](https://github.com/NVIDIA/TensorRT-LLM/blob/feat/reflection-beam/docs/source/deployment-guide/deployment-guide-for-beam-on-trtllm.md) | No container published yet — build from the branch sources per the guide | + ## Model-Feature Support Matrix (Key Models) Note: Support for other models may vary. Features marked "N/A" are not applicable to the model architecture. From a4287e0f181c6bc1106dd518cfb88031678b3dce Mon Sep 17 00:00:00 2001 From: Brian Nguyen Date: Fri, 30 Oct 2026 14:40:46 -0500 Subject: [PATCH 2/2] [None][doc] Rename the side-branch section per review Signed-off-by: Brian Nguyen --- docs/source/models/supported-models.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/source/models/supported-models.md b/docs/source/models/supported-models.md index 1afacf769864..58aeea645067 100644 --- a/docs/source/models/supported-models.md +++ b/docs/source/models/supported-models.md @@ -57,7 +57,7 @@ The following is a table of supported models for the PyTorch backend: | `WhisperForConditionalGeneration` | Whisper | `openai/whisper-large-v3` | -## Models Supported on Day-0 Side Branches +## Models Supported outside main branch Some newly released models are first supported on a dedicated feature branch, with release containers published from that branch, before the support is merged into `main`. The entries