Repository Issue Activity (beta)

#8071 All output buffers not written to

#8065 [LiteRT.js / WebGPU] CompiledModel produces correct output only on its first inference on Adreno 8xx (Galaxy Z Fold 7, Chrome 149)

#7005 How to add NPU support to .literlm format model?

#8062 [Google Tensor] Generative decoder LLMs (Gemma-3-1B, Gemma-4, Qwen2.5) fail to compile to Tensor_G5 with INTERNAL; vision compiles in the same env

#7598 [Bug] AOT compilation succeeds for MediaPipe selfie model but fails for YOLO11n on Tensor G4 — INTERNAL error at compiler_plugin.cc:481

#7629 Model compilation failed for Pixel 10 Pro on Ubuntu 20.04 via Google Tensor SDK Beta for NPU inference

#7997 Rust segmentation example fails to build and run on macOS (arm64)

#7878 [Intel-OpenVINO] The `openvino_compiler_plugin_test` and `dispatch_api_openvino_test` targets cannot be built

#8037 [LiteRT] CompiledModel.create() fails to trigger on-device JIT for float models, and host-side AOT compiler ignores --soc_model (hardcoded to v79)

#7996 Expose Custom OP Dispatcher to Python API

#7459 LiteRtLayout struct has different binary layout under MSVC vs GCC/Clang (non-portable bit-field packing)

#7858 LiteRT Next CompiledModel + Accelerator.NPU (Qualcomm HTP) ~6x slower than the Interpreter + QNN TFLite Delegate path for the same INT8 model; QualcommOptions have no measurable effect

#7891 CompiledModel: NPU silently falls back to CPU with no diagnostic for custom models (S23 Ultra / Pixel 9 Pro)

#7936 EfficientDet fails on Android GPU (S20/S23) with CompiledModel

#7693 EmbeddingGemma 300M returns all-zero vectors on Mac GPU unless precision is forced to FP32

#7787 Tensor SDK Beta — Dispatcher / runtime protocol mismatch makes AOT path unusable on Pixel 10 Pro XL

#7603 [MediaTek] Per-channel quant rejected on MT6989/6991/6993 AOT

#7713 [Feature Request] Native support for 1-bit and Ternary (1.58-bit) quantization

#7602 LiteRtCreateEnvironment silently disables NPU when kLiteRtEnvOptionTagDispatchLibraryDir is missing (dynamic-link build)

#7636 buffer moved twice

#7216 Rockchip RK3588 series NPU

#7532 LiteRT.js / WebNN GPU: Unexpected error for litert-community/xcit_tiny_24_p8_224

#125 Make TensorFlow Lite available as Swift Package Manager package

#7470 No Prebuilt iOS Library for LiteRT Next

#5499 Guidance on using FastVLM with NPU via LiteRT/LiteRT-LM