All issues

Search results

Stop training worker immediately when cancelled during compile

Inference technologies

Shield

Status: Open.

#5218 In tenstorrent/tt-inference-server;

· ztorlakTT opened on Sep 23, 2026

Release Llama-3.1-8B-Instruct QB2 in tt-inference-server

community

model_readiness_support

Status: Open.

#5209 In tenstorrent/tt-inference-server;

· yieldthought opened on Sep 22, 2026

Host memory leak in vLLM EngineCore under fixed load in images built from tt-metal before 2026-07-20; likely Tracy client buffering (tt-metal#49766, fixed by tt-metal#50480)

bugSomething isn't workingSomething isn't working

Status: Open.

Bug

#5198 In tenstorrent/tt-inference-server;

· tstescoTT opened on Sep 22, 2026

Change the numbers in TTS scheduler according to Customer's AI needs

Inference technologies

Llama 8B

llm-engine

Shield

Status: Open.

#5186 In tenstorrent/tt-inference-server;

· nanicicTT opened on Sep 18, 2026

[Helm] Decide the boundary with the tt-operator multi-host inference chart

Helm

tt-operator

Status: Open.

Feature

#5178 In tenstorrent/tt-inference-server;

· sott0n opened on Sep 17, 2026

[Helm] Extend the nightly hardware smoke matrix beyond a single n150

Helm

tt-operator

Status: Open.

Feature

#5177 In tenstorrent/tt-inference-server;

· sott0n opened on Sep 17, 2026

[Helm v0.3] Publish the chart as an OCI artifact on GHCR

Helm

tt-operator

Status: Open.

Feature

#5176 In tenstorrent/tt-inference-server;

· sott0n opened on Sep 17, 2026

[Helm v0.3] device × app telemetry correlation — join design + mapping exporter

Helm

tt-operator

Status: Open.

Feature

#5175 In tenstorrent/tt-inference-server;

· sott0n opened on Sep 17, 2026

·  

tt-operator-v0.4

[Helm v0.3] device=gpu renders a Deployment with no accelerator request

Helm

tt-operator

Status: Open.

Feature

#5174 In tenstorrent/tt-inference-server;

· sott0n opened on Sep 17, 2026

·  

tt-operator-v0.4

[Helm v0.3] Supported image-release floor (vLLM 0.11.0+, media/forge NO_AUTH 0.15.0+) + appVersion provenance

Helm

tt-operator

Status: Open.

Feature

#5173 In tenstorrent/tt-inference-server;

· sott0n opened on Sep 17, 2026

·  

tt-operator-v0.4

[Helm v0.3] tt-operator v0.3.0 compatibility + supported-version matrix

Helm

tt-operator

Status: Open.

Feature

#5172 In tenstorrent/tt-inference-server;

· sott0n opened on Sep 17, 2026

·  

tt-operator-v0.4

[tt cli report] tt serve fails to autodetect available cards

community

Status: Open.

#5158 In tenstorrent/tt-inference-server;

· chathaway-codes opened on Sep 15, 2026