Infinity GitHub repository · inference-research

Fields

id

sources/niche-infinity-repo

URL

https://github.com/michaelfeil/infinity

Publisher

Michael Feil

Kind

repo

Accessed

2026-10-03

notes

Infinity README: embedding/reranking/multimodal serving engine layered on three backends (PyTorch, optimum for ONNX/TensorRT, CTranslate2). Documents CUDA, ROCm (MI200/MI300) and CPU docker images; int8 support is experimental on cpu/cuda and fp8 is experimental on H100/MI300 only. No lora adapters, merged models only for the ColBERT/ColPali-family paths. API metadata 2026-10-03: created 2023-10-11, last push 2026-03-24, last release 0.0.77 (2025-08-22) so the release cadence has slowed, MIT, 281 open issues, 2949 stars. UNVERIFIED: the claim that Infinity was 'acquired by / merged into Milvus' could not be confirmed - only 15 code references to infinity inside milvus-io/milvus and 2 issue mentions.

Notes

Infinity README: embedding/reranking/multimodal serving engine layered on three backends (PyTorch, optimum for ONNX/TensorRT, CTranslate2). Documents CUDA, ROCm (MI200/MI300) and CPU docker images; int8 support is experimental on cpu/cuda and fp8 is experimental on H100/MI300 only. No lora adapters, merged models only for the ColBERT/ColPali-family paths. API metadata 2026-10-03: created 2023-10-11, last push 2026-03-24, last release 0.0.77 (2025-08-22) so the release cadence has slowed, MIT, 281 open issues, 2949 stars. UNVERIFIED: the claim that Infinity was 'acquired by / merged into Milvus' could not be confirmed - only 15 code references to infinity inside milvus-io/milvus and 2 issue mentions.

Cross-references

This record references no other records.