vllm-tpu
A high-throughput and memory-efficient inference and serving engine for LLMs
Decision gist · record as of 2026-08-14
Yes, if you are deploying LLMs on Google TPUs and need production-grade serving infrastructure. The active maintenance, permissive license, and broad model support make it a solid choice. Medium install friction is acceptable for a specialized inference engine. No security vulnerabilities reported. Not suitable if you lack TPU hardware or require GPU-only solutions.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- Requires Python 3.10–3.14; TPU hardware or compatible accelerator needed for actual inference; large model weights must be downloaded from Hugging Face.
- Medium install friction due to 66 runtime dependencies including transformers, fastapi, and specialized inference libraries.
- Active maintenance with recent releases (14 days old); repository shows strong community engagement with 89063 stars and ongoing development.
License · maintenance · safety
Apache-2.0 (permissive) — Apache-2.0 permissive license allows commercial use, modification, and distribution with minimal restrictions, making it suitable for production deployments and proprietary applications.
last release 2026-07-31 (14 days) · last repo commit 2026-08-14 · 89,063 stars
0 known vulnerabilities (OSV.dev, 2026-08-14) · 85,080 downloads/mo, #13,954 on PyPI
Alternatives
Verify before relying
pip install vllm-tpu
from vllm import LLM, SamplingParams
llm = LLM(model="meta-llama/Llama-2-hf")
outputs = llm.generate(["Hello, my name is"], SamplingParams(temperature=0.8))- Specific performance benchmarks for TPU vs. GPU inference on common models
- Whether the 0.26.0 release includes all features from the main vLLM repository or is TPU-specific subset
- Memory requirements and minimum TPU configuration for serving production models
What it is and what it does
vllm-tpu is a specialized distribution of vLLM optimized for serving large language models on Google TPUs. It provides high-throughput inference with efficient memory management through PagedAttention, continuous batching, and support for 200+ model architectures including decoder-only LLMs, mixture-of-experts models, and multi-modal models. The package includes an OpenAI-compatible API server, support for structured output generation via xgrammar and guidance, and distributed inference across multiple TPUs.
The package brings together 66 runtime dependencies—including transformers, fastapi, tokenizers, and specialized inference libraries—to handle the full stack of model loading, tokenization, inference scheduling, and serving. It targets developers and researchers deploying LLMs at scale, offering both programmatic access via Python and a production-ready HTTP API. Installation requires Python 3.10–3.14 and TPU hardware or compatible accelerators.
Use it for
- Deploy open-source LLMs on Google TPUs with high throughput for production inference
- Build OpenAI-compatible API services for LLM inference without modifying client code
- Serve multi-modal models or mixture-of-experts models on TPU clusters
- Generate structured outputs using xgrammar or guidance constraints during inference
- Implement distributed inference across multiple TPUs using tensor, pipeline, or data parallelism
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes, if you are deploying LLMs on Google TPUs and need production-grade serving infrastructure.
The active maintenance, permissive license, and broad model support make it a solid choice. Medium install friction is acceptable for a specialized inference engine. No security vulnerabilities reported. Not suitable if you lack TPU hardware or require GPU-only solutions.
Install
vllm-tpu on PyPI
Before you install
Medium install friction due to 66 runtime dependencies including transformers, fastapi, and specialized inference libraries. Active maintenance with recent releases (14 days old); repository shows strong community engagement with 89063 stars and ongoing development.
Requires Python 3.10–3.14; TPU hardware or compatible accelerator needed for actual inference; large model weights must be downloaded from Hugging Face.
License in practice
Apache-2.0 permissive license allows commercial use, modification, and distribution with minimal restrictions, making it suitable for production deployments and proprietary applications.
Quickstart
pip install vllm-tpu
from vllm import LLM, SamplingParams
llm = LLM(model="meta-llama/Llama-2-hf")
outputs = llm.generate(["Hello, my name is"], SamplingParams(temperature=0.8))Verify before relying
- Specific performance benchmarks for TPU vs. GPU inference on common models
- Whether the 0.26.0 release includes all features from the main vLLM repository or is TPU-specific subset
- Memory requirements and minimum TPU configuration for serving production models
Package facts
| License | Apache-2.0 permissive |
| Python support | Supports the current Python release <3.15,>=3.10 |
| Install friction | Medium. Platform-specific wheel |
| Runtime dependencies | 66 packagesregexcachetoolspsutilsentencepiecenumpyrequeststqdmblake3py-cpuinfotransformerstokenizerssafetensorsprotobuffastapistarletteaiohttpopenaipydanticprometheus_clientpillowprometheus-fastapi-instrumentatortiktokenlm-format-enforcerllguidanceoutlines_corelarkxgrammartyping_extensionsfilelockpartial-json-parser |
| Maintenance | Actively maintained 14 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 85,080 / month, #13,954 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
| Classifiers | Intended Audience :: DevelopersIntended Audience :: Information TechnologyIntended Audience :: Science/ResearchProgramming Language :: Python :: 3.10Programming Language :: Python :: 3.11Programming Language :: Python :: 3.12Programming Language :: Python :: 3.13Programming Language :: Python :: 3.14Topic :: Scientific/Engineering :: Artificial IntelligenceTopic :: Scientific/Engineering :: Information Analysis |
Evidence: vllm_tpu-0.26.0-cp312-cp312-manylinux_2_24_x86_64.whl
Tags
Let your AI agent find packages like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 14,416 PyPI packages by what they can do, searchable in plain language.
wish › “LLM inference serving”
- vllm-tpuvllm-tpu is a high-throughput LLM inference and serving engine…
- vllmvLLM is a high-throughput inference and serving engine for large…
- sglangSGLang is a serving framework that runs large language models and…
Give your agent the search over MCP, or paste the wish link into any chat.
More Artificial Intelligence packages
LiteLLM provides a unified Python interface to call 100+ LLM providers (OpenAI, Anthropic, Gemini, Bedrock, Azure, and others) using OpenAI-compatible API format, available as both a Python SDK and a self-hosted AI Gateway proxy server.
Install it if you need to work with multiple LLM providers or want to centralize LLM routing in your organization.
Client library and CLI tool for downloading, uploading, and managing models, datasets, and repositories on the Hugging Face Hub platform.
Install it if you work with Hugging Face Hub models or datasets.
LangChain provides a framework for building agents and LLM-powered applications by composing language models, tools, and memory through a unified API that abstracts over multiple model providers.
hf-xet provides chunk-based deduplication and efficient file transfer for the Hugging Face Hub, enabling faster uploads and downloads of large files with local disk caching.
Tokenizers converts raw text into token sequences for NLP models, with support for training custom vocabularies and using pre-built tokenizers (BPE, WordPiece) optimized for speed via Rust.
Transformers provides a unified framework for loading, fine-tuning, and running state-of-the-art pretrained models across text, vision, audio, video, and multimodal tasks using PyTorch, JAX, or TensorFlow.
Install it if you need to run or train any transformer-based model for NLP, vision, audio, or multimodal tasks.
See also tpu-inference · vllm · vllm-cpu · sglang · lmcache · mooncake-transfer-engine · llmcompressor · vllm-router · smg-grpc-servicer · ipex-llm