TensorRT-LLM

SkillCloud & infra

This skill guides your AI to set up and run TensorRT-LLM, NVIDIA's toolkit for running language models faster. Once added, your AI can help you get quicker model responses and higher throughput on NVIDIA GPUs, including setup for production use on A100 or H100 cards.

Available today. Use it from your connected AI after setup.

No other account needed.

After adding the skill, ask your AI to set up TensorRT-LLM on your NVIDIA GPU. It will walk through the setup steps and get a model running with faster inference.

Then ask your AI: use the TensorRT-LLM skill

What your AI can do with it

Set up TensorRT-LLM for faster model inference on NVIDIA A100 or H100 GPUs
Run language models up to 10-100x faster than PyTorch
Apply FP8 or INT4 quantization to reduce latency and memory use
Handle many requests at once with in-flight batching
Spread model workloads across multiple GPUs for more throughput

What this skill tells your AI

The instructions your AI receives, as published by openlair/dr-claw in skills/inference-serving/tensorrt-llm/SKILL.md and read by ahel’s review.

NVIDIA's open-source library for optimizing LLM inference with state-of-the-art performance on NVIDIA GPUs.

When to use TensorRT-LLM

Use TensorRT-LLM when:

Deploying on NVIDIA GPUs (A100, H100, GB200)
Need maximum throughput (24,000+ tokens/sec on Llama 3)
Require low latency for real-time applications
Working with quantized models (FP8, INT4, FP4)
Scaling across multiple GPUs or nodes

Use vLLM instead when:

Need simpler setup and Python-first API
Want PagedAttention without TensorRT compilation
Working with AMD GPUs or non-NVIDIA hardware

Use llama.cpp instead when:

Deploying on CPU or Apple Silicon
Need edge deployment without NVIDIA GPUs
Want simpler GGUF quantization format

Quick start

Installation

# Docker (recommended)
docker pull nvidia/tensorrt_llm:latest

# pip install
pip install tensorrt_llm==1.2.0rc3

# Requires CUDA 13.0.0, TensorRT 10.13.2, Python 3.10-3.12

Basic inference

from tensorrt_llm import LLM, SamplingParams

# Initialize model
llm = LLM(model="meta-llama/Meta-Llama-3-8B")

# Configure sampling
sampling_params = SamplingParams(
    max_tokens=100,
    temperature=0.7,
    top_p=0.9
)

# Generate
prompts = ["Explain quantum computing"]
outputs = llm.generate(prompts, sampling_params)

for output in outputs:
    print(output.text)

Serving with trtllm-serve

# Start server (automatic model download and compilation)
trtllm-serve meta-llama/Meta-Llama-3-8B \
    --tp_size 4 \              # Tensor parallelism (4 GPUs)
    --max_batch_size 256 \
    --max_num_tokens 4096

# Client request
curl -X POST http://localhost:8000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "meta-llama/Meta-Llama-3-8B",
    "messages": [{"role": "user", "content": "Hello!"}],
    "temperature": 0.7,
    "max_tokens": 100
  }'

Key features

Performance optimizations

In-flight batching: Dynamic batching during generation
Paged KV cache: Efficient memory management
Flash Attention: Optimized attention kernels
Quantization: FP8, INT4, FP4 for 2-4× faster inference
CUDA graphs: Reduced kernel launch overhead

Parallelism

Tensor parallelism (TP): Split model across GPUs
Pipeline parallelism (PP): Layer-wise distribution
Expert parallelism: For Mixture-of-Experts models
Multi-node: Scale beyond single machine

Advanced features

Speculative decoding: Faster generation with draft models
LoRA serving: Efficient multi-adapter deployment
Disaggregated serving: Separate prefill and generation

Common patterns

Quantized model (FP8)

from tensorrt_llm import LLM

# Load FP8 quantized model (2× faster, 50% memory)
llm = LLM(
    model="meta-llama/Meta-Llama-3-70B",
    dtype="fp8",
    max_num_tokens=8192
)

# Inference same as before
outputs = llm.generate(["Summarize this article..."])

Multi-GPU deployment

# Tensor parallelism across 8 GPUs
llm = LLM(
    model="meta-llama/Meta-Llama-3-405B",
    tensor_parallel_size=8,
    dtype="fp8"
)

Batch inference

# Process 100 prompts efficiently
prompts = [f"Question {i}: ..." for i in range(100)]

outputs = llm.generate(
    prompts,
    sampling_params=SamplingParams(max_tokens=200)
)

# Automatic in-flight batching for maximum throughput