Search results

[None][infra] Add Docker BuildKit cache (--cache-to/--cache-from) to reduce Docker image build times

Infra<NV>automated tests, build checks, github actions, system stability & efficiency.<NV>automated tests, build checks, github actions, system stability & efficiency.

Status: Open.

#13023 In NVIDIA/TensorRT-LLM;

· sandeshwani opened on Apr 14, 2026

Add concurrency control and least-privilege permissions to PR check workflow

Infra<NV>automated tests, build checks, github actions, system stability & efficiency.<NV>automated tests, build checks, github actions, system stability & efficiency.

Status: Open.

#13021 In NVIDIA/TensorRT-LLM;

· sandeshwani opened on Apr 14, 2026

[None][fix] multi_stream_moe + MLIR accuracy regression in monolithic CUDA graph decode path

AutoDeploy<NV> AutoDeploy Backend<NV> AutoDeploy Backend

CUDA Graph

Customized kernels<NV>Specialized/modified CUDA kernels in TRTLLM for LLM ops, beyond standard TRT. Dev & perf.<NV>Specialized/modified CUDA kernels in TRTLLM for LLM ops, beyond standard TRT. Dev & perf.

Status: Open.

#12954 In NVIDIA/TensorRT-LLM;

· suyoggupta opened on Apr 11, 2026

[AutoDeploy][Bug]: verify accuracy tests on sharded models

AutoDeploy<NV> AutoDeploy Backend<NV> AutoDeploy Backend

feature requestNew feature or request. This includes new model, dtype, functionality supportNew feature or request. This includes new model, dtype, functionality support

Status: Open.

#12951 In NVIDIA/TensorRT-LLM;

· greg-kwasniewski1 opened on Apr 11, 2026

[Feature]: [AutoDeploy] Accuracy issue w/ piecewise cudagraph for DeepSeek R1

CUDA Graph

feature requestNew feature or request. This includes new model, dtype, functionality supportNew feature or request. This includes new model, dtype, functionality support

Status: Open.

Task

#12949 In NVIDIA/TensorRT-LLM;

· taylor-yb-lee opened on Apr 11, 2026

[Usage]: Observing very less throughput-[04/10/2026-11:43:27] [TRT-LLM] [E] Streaming: Failed to process token batch of 10 tokens for request 2

Pytorch<NV>Pytorch backend related issues<NV>Pytorch backend related issues

questionFurther information is requestedFurther information is requested

triagedIssue has been triaged by maintainersIssue has been triaged by maintainers

Status: Open.

#12933 In NVIDIA/TensorRT-LLM;

· sravan500 opened on Apr 10, 2026

[Feature]: Direct model loading from object storage with Runai Model Streamer

feature requestNew feature or request. This includes new model, dtype, functionality supportNew feature or request. This includes new model, dtype, functionality support

Status: Open.

#12900 In NVIDIA/TensorRT-LLM;

· GuyStone opened on Apr 9, 2026

[Bug]: Garbage output with modelopt FP8 checkpoint in 1.2.0: use_fp8_context_fmha forced True, regression from 1.1.0

bugSomething isn't workingSomething isn't working

Model optimization<NV>Model-specific performance optimizations and tuning<NV>Model-specific performance optimizations and tuning

Status: Open.

#12877 In NVIDIA/TensorRT-LLM;

· langzhao-netizen opened on Apr 9, 2026

[RFC]: ForwardPassMetrics — per-iteration scheduler telemetry for dynamo integration

RFC

Status: Open.

#12854 In NVIDIA/TensorRT-LLM;

· venkywonka opened on Apr 8, 2026

[PyTorch backend][Qwen3] embed_tokens ignores allreduce_strategy and falls back to AUTO

Pytorch<NV>Pytorch backend related issues<NV>Pytorch backend related issues

triagedIssue has been triaged by maintainersIssue has been triaged by maintainers

Status: Open.

#12840 In NVIDIA/TensorRT-LLM;

· CarstyYou opened on Apr 8, 2026

[New Model]: Support Cohere Labs Command A Vision

new modelRequest to add a new modelRequest to add a new model

Status: Open.

#12839 In NVIDIA/TensorRT-LLM;

· torotoki opened on Apr 8, 2026

[Bug]: trtllm-serve 1.3.0rc10 returns "Unknown modality: image" for Qwen2.5-VL-7B-Instruct-NVFP4 on DGX Spark

bugSomething isn't workingSomething isn't working

MultimodalLabel for issues & PRs regarding Multimodal related objectsLabel for issues & PRs regarding Multimodal related objects

Status: Open.

#12824 In NVIDIA/TensorRT-LLM;

· xj999 opened on Apr 8, 2026