Ross ROSS = Recommend OSS · open-source software intelligence for agents

triton-inference-server/server

The Triton Inference Server provides an optimized cloud and edge inferencing solution. observed · 2026-08-28

github.com/triton-inference-server/server · homepage · Python · BSD-3-Clause (permissive) observed · 2026-08-28

Health v2 · maintenance only

98/100

  • Activity 99
  • Release rhythm 95
  • Longevity 100
How is this computed?

round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10) — computed 2026-09-03. Adoption (stars, forks) is never an input.

  • gap_med: 28.0
  • age_days: 2890
  • days_rel: 35
  • days_push: 7
  • n_releases_24m: 23

Full methodology

Adoption not part of the score

10939 stars · 1830 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

NVIDIA Triton Inference Server is an open-source inference serving software that deploys AI models from multiple frameworks (TensorRT, PyTorch, ONNX, OpenVINO, Python, vLLM, and more) behind HTTP/REST and gRPC APIs. It supports cloud, data center, edge, and embedded deployments on NVIDIA GPUs, x86/ARM CPUs, and AWS Inferentia, with dynamic batching, ensembles, and streaming inference.

Use cases

  • serve pytorch models in production with a rest api
  • deploy llm like llama2 or phi-3 with vllm or tensorrt-llm backend
  • run batched gpu inference for many models from one server
  • host onnx models for real-time inference at the edge
  • serve models over grpc with dynamic batching and ensembles
  • deploy models on nvidia gpus in docker containers
  • stream audio or video inference requests to a model server

When to choose

  • you need production-grade, high-throughput model serving on NVIDIA GPUs
  • you must serve models from multiple frameworks through a single API
  • you want dynamic batching, model ensembles, and streaming out of the box
  • you deploy across cloud, data center, and edge with the same tooling

When to avoid

  • you only need simple CPU-only inference for a single small model
  • you want a lightweight embedded inference runtime rather than a server
  • your stack is tightly tied to a non-NVIDIA accelerator without supported backends

Facets

service · maturity stable

llm-inference machine-learning deep-learning http-server rpc gpu-computing monitoring machine-learning deep-learning large-language-models cloud-computing developer-tools self-hosted cloud self-hosted cross-platform inference-server model-serving nvidia tensorrt pytorch onnx batching grpc edge-deployment model-repository linux docker gpu

10 sources

Member repositories

RepositoryRoleHealth v2
triton-inference-server/servermain98

For agents

markdown · JSON · MCP: product_card(name="triton-inference-server/server")

Data as of 2026-08-30T08:39:29.467469+00:00 · Report a problem