Search results

[Feature]: Fix Qwen3.5, Gemma4 CI

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

Testing<NV>Continuous integration, build system, and testing infrastructure issues<NV>Continuous integration, build system, and testing infrastructure issues

Status: Open.

#13390 In NVIDIA/TensorRT-LLM;

· bmarimuthu-nv opened on Apr 23, 2026

[Feature]: Minimax 2.7 Perf Improvements

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

Status: Open.

#13389 In NVIDIA/TensorRT-LLM;

· bmarimuthu-nv opened on Apr 23, 2026

[Feature]: Support Nemotron Nano Omni

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

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

Status: Open.

#13388 In NVIDIA/TensorRT-LLM;

· bmarimuthu-nv opened on Apr 23, 2026

[Usage]: Is Qwen3.5 MTP currently supported in TensorRT-LLM?

questionFurther information is requestedFurther information is requested

Speculative Decoding<NV>MTP/Eagle/Medusa/Lookahead/Prompt-Lookup-Decoding/Draft-Target-Model/ReDrafter<NV>MTP/Eagle/Medusa/Lookahead/Prompt-Lookup-Decoding/Draft-Target-Model/ReDrafter

Status: Open.

#13361 In NVIDIA/TensorRT-LLM;

· minedec opened on Apr 23, 2026

[Feature]: Enhance AllToAll performance

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.

Task

#13360 In NVIDIA/TensorRT-LLM;

· taylor-yb-lee opened on Apr 23, 2026

[Feature]: AutoDeploy: deprecate export_to_onnx backend

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.

Task

#13353 In NVIDIA/TensorRT-LLM;

· lucaslie opened on Apr 22, 2026

[AutoDeploy][spec-dec] resize_kv_cache sample batch classified as decode-only when max_num_tokens // max_batch_size == 1

Speculative Decoding<NV>MTP/Eagle/Medusa/Lookahead/Prompt-Lookup-Decoding/Draft-Target-Model/ReDrafter<NV>MTP/Eagle/Medusa/Lookahead/Prompt-Lookup-Decoding/Draft-Target-Model/ReDrafter

Status: Open.

#13348 In NVIDIA/TensorRT-LLM;

· govind-ramnarayan opened on Apr 22, 2026

[Bug]: Re-enable piecewise cudagraph for Qwen3.5

bugSomething isn't workingSomething isn't working

CUDA Graph

Status: Open.

#13321 In NVIDIA/TensorRT-LLM;

· bmarimuthu-nv opened on Apr 22, 2026

[Bug]: kv_cache_connector: decode-continuation blocks never allocated — GenerationRequest's mCacheBlockIds frozen while mNumTokens grows

bugSomething isn't workingSomething isn't working

KV-Cache Managementkv-cache management for efficient LLM inferencekv-cache management for efficient LLM inference

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

Status: Open.

#13320 In NVIDIA/TensorRT-LLM;

· GavinZhu-GMI opened on Apr 22, 2026

[Bug]: [Bug] DeepSeek-V3 MoE init fails with "Not all rank allocating same size" when using NVLinkOneSided (mpirun, ep_size=8)

bugSomething isn't workingSomething isn't working

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

Scale-out<NV>Multi-GPU and distributed inference scaling issues, tensor/pipeline/data parallelism<NV>Multi-GPU and distributed inference scaling issues, tensor/pipeline/data parallelism

Status: Open.

#13319 In NVIDIA/TensorRT-LLM;

· liyandong001 opened on Apr 22, 2026

[Bug]: Scheduler deadlock on main + #12976 + #13029: AssertionError total_num_tokens > max_num_tokens in _prepare_tp_inputs under KV offload + chunked prefill permanently hangs the event loop

bugSomething isn't workingSomething isn't working

KV-Cache Managementkv-cache management for efficient LLM inferencekv-cache management for efficient LLM inference

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

Status: Open.

#13318 In NVIDIA/TensorRT-LLM;

· Saddss opened on Apr 22, 2026

[Bug]: Multimodal example code crashes for Gemma3 with AttributeError related to kv_layout

bugSomething isn't workingSomething isn't working

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

Status: Open.

#13293 In NVIDIA/TensorRT-LLM;

· torotoki opened on Apr 21, 2026