---
name: serving-llms-vllm-57eaf97e982f
description: Serves LLMs with high throughput using vLLM's PagedAttention and continuous batching. Use when deploying production LLM APIs, optimizing inference latency/throughput, or serving models with limited GPU memory. Supports OpenAI-compatible endpoints, quantization (GPTQ/AWQ/FP8), and tensor parallelism.
version: 1.0.0
author: Orchestra Research
license: MIT
tags: [vLLM, Inference Serving, PagedAttention, Continuous Batching, High Throughput, Production, OpenAI API, Quantization, Tensor Parallelism]
dependencies: [vllm, torch, transformers]
---

# vLLM - High-Performance LLM Serving

## Quick start

vLLM achieves 24x higher throughput than standard transformers through PagedAttention (block-based KV cache) and continuous batching (mixing prefill/decode requests).

**Installation**:
```bash
pip install vllm
```

**Basic offline inference**:
```python
from vllm import LLM, SamplingParams

llm = LLM(model="meta-llama/Llama-3-8B-Instruct")
sampling = SamplingParams(temperature=0.7, max_tokens=256)

outputs = llm.generate(["Explain quantum computing"], sampling)
print(outputs[0].outputs[0].text)
```

**OpenAI-compatible server**:
```bash
vllm serve meta-llama/Llama-3-8B-Instruct

# Query with OpenAI SDK
python -c "
from openai import OpenAI
client = OpenAI(base_url='http://localhost:8000/v1', api_key='EMPTY')
print(client.chat.completions.create(
    model='meta-llama/Llama-3-8B-Instruct',
    messages=[{'role': 'user', 'content': 'Hello!'}]
).choices[0].message.content)
"
```

## Common workflows

### Workflow 1: Production API deployment

Copy this checklist and track progress:

```
Deployment Progress:
- [ ] Step 1: Configure server settings
- [ ] Step 2: Test with limited traffic
- [ ] Step 3: Enable monitoring
- [ ] Step 4: Deploy to production
- [ ] Step 5: Verify performance metrics
```

**Step 1: Configure server settings**

Choose configuration based on your model size:

```bash
# For 7B-13B models on single GPU
vllm serve meta-llama/Llama-3-8B-Instruct \
  --gpu-memory-utilization 0.9 \
  --max-model-len 8192 \
  --port 8000

# For 30B-70B models with tensor parallelism
vllm serve meta-llama/Llama-2-70b-hf \
  --tensor-parallel-size 4 \
  --gpu-memory-utilization 0.9 \
  --quantization awq \
  --port 8000

# For production with caching and metrics
vllm serve meta-llama/Llama-3-8B-Instruct \
  --gpu-memory-utilization 0.9 \
  --enable-prefix-caching \
  --enable-metrics \
  --metrics-port 9090 \
  --port 8000 \
  --host 0.0.0.0
```

**Step 2: Test with limited traffic**

Run load test before production:

```bash
# Install load testing tool
pip install locust

# Create test_load.py with sample requests
# Run: locust -f test_load.py --host http://localhost:8000
```

Verify TTFT (time to first token) < 500ms and throughput > 100 req/sec.

**Step 3: Enable monitoring**

vLLM exposes Prometheus metrics on port 9090:

```bash
curl http://localhost:9090/metrics | grep vllm
```

Key metrics to monitor:
- `vllm:time_to_first_token_seconds` - Latency
- `vllm:num_requests_running` - Active requests
- `vllm:gpu_cache_usage_perc` - KV cache utilization

**Step 4: Deploy to production**

Use Docker for consistent deployment:

```bash
# Run vLLM in Docker
docker run --gpus all -p 8000:8000 \
  vllm/vllm-openai:latest \
  --model meta-llama/Llama-3-8B-Instruct \
  --gpu-memory-utilization 0.9 \
  --enable-prefix-caching
```

**Step 5: Verify performance metrics**

Check that deployment meets targets:
- TTFT < 500ms (for short prompts)
- Throughput > target req/sec
- GPU utilization > 80%
- No OOM errors in logs

### Workflow 2: Offline batch inference

For processing large datasets without server overhead.

Copy this checklist:

```
Batch Processing:
- [ ] Step 1: Prepare input data
- [ ] Step 2: Configure LLM engine
- [ ] Step 3: Run batch inference
- [ ] Step 4: Process results
```

**Step 1: Prepare input data**

```python
# Load prompts from file
prompts = []
with open("prompts.txt") as f:
    prompts = [line.strip() for line in f]

print(f"Loaded {len(prompts)} prompts")
```

**Step 2: Configure LLM engine**

```python
from vllm import LLM, SamplingParams

llm = LLM(
    model="meta-llama/Llama-3-8B-Instruct",
    tensor_parallel_size=2,  # Use 2 GPUs
    gpu_memory_utilization=0.9,
    max_model_len=4096
)

sampling = SamplingParams(
    temperature=0.7,
    top_p=0.95,
    max_tokens=512,
    stop=["</s>", "\n\n"]
)
```

**Step 3: Run batch inference**

vLLM automatically batches requests for efficiency:

```python
# Process all prompts in one call
outputs = llm.generate(prompts, sampling)

# vLLM handles batching internally
# No need to manually chunk prompts
```

**Step 4: Process results**

```python
# Extract generated text
results = []
for output in outputs:
    prompt = output.prompt
    generated = output.outputs[0].text
    results.append({
        "prompt": prompt,
        "generated": generated,
        "tokens": len(output.outputs[0].token_ids)
    })

# Save to file
import json
with open("results.jsonl", "w") as f:
    for result in results:
        f.write(json.dumps(result) + "\n")

print(f"Processed {len(results)} prompts")
```

### Workflow 3: Quantized model serving

Fit large models in limited GPU memory.

```
Quantization Setup:
- [ ] Step 1: Choose quantization method
- [ ] Step 2: Find or create quantized model
- [ ] Step 3: Launch with quantization flag
- [ ] Step 4: Verify accuracy
```

**Step 1: Choose quantization method**

- **AWQ**: Best for 70B models, minimal accuracy loss
- **GPTQ**: Wide model support, good compression
- **FP8**: Fastest on H100 GPUs

**Step 2: Find or create quantized model**

Use pre-quantized models from HuggingFace:

```bash
# Search for AWQ models
# Example: TheBloke/Llama-2-70B-AWQ
```

**Step 3: Launch with quantization flag**

```bash
# Using pre-quantized model
vllm serve TheBloke/Llama-2-70B-AWQ \
  --quantization awq \
  --tensor-parallel-size 1 \
  --gpu-memory-utilization 0.95

# Results: 70B model in ~40GB VRAM
```

**Step 4: Verify accuracy**

Test outputs match expected quality:

```python
# Compare quantized vs non-quantized responses
# Verify task-specific performance unchanged
```

## When to use vs alternatives

**Use vLLM when:**
- Deploying production LLM APIs (100+ req/sec)
- Serving OpenAI-compatible endpoints
- Limited GPU memory but need large models
- Multi-user applications (chatbots, assistants)
- Need low latency with high throughput

**Use alternatives instead:**
- **llama.cpp**: CPU/edge inference, single-user
- **HuggingFace transformers**: Research, prototyping, one-off generation
- **TensorRT-LLM**: NVIDIA-only, need absolute maximum performance
- **Text-Generation-Inference**: Already in HuggingFace ecosystem

## Common issues

**Issue: Out of memory during model loading**

Reduce memory usage:
```bash
vllm serve MODEL \
  --gpu-memory-utilization 0.7 \
  --max-model-len 4096
```

Or use quantization:
```bash
vllm serve MODEL --quantization awq
```

**Issue: Slow first token (TTFT > 1 second)**

Enable prefix caching for repeated prompts:
```bash
vllm serve MODEL --enable-prefix-caching
```

For long prompts, enable chunked prefill:
```bash
vllm serve MODEL --enable-chunked-prefill
```

**Issue: Model not found error**

Use `--trust-remote-code` for custom models:
```bash
vllm serve MODEL --trust-remote-code
```

**Issue: Low throughput (<50 req/sec)**

Increase concurrent sequences:
```bash
vllm serve MODEL --max-num-seqs 512
```

Check GPU utilization with `nvidia-smi` - should be >80%.

**Issue: Inference slower than expected**

Verify tensor parallelism uses power of 2 GPUs:
```bash
vllm serve MODEL --tensor-parallel-size 4  # Not 3
```

Enable speculative decoding for faster generation:
```bash
vllm serve MODEL --speculative-model DRAFT_MODEL
```

## Advanced topics

**Server deployment patterns**: See [references/server-deployment.md](references/server-deployment.md) for Docker, Kubernetes, and load balancing configurations.

**Performance optimization**: See [references/optimization.md](references/optimization.md) for PagedAttention tuning, continuous batching details, and benchmark results.

**Quantization guide**: See [references/quantization.md](references/quantization.md) for AWQ/GPTQ/FP8 setup, model preparation, and accuracy comparisons.

**Troubleshooting**: See [references/troubleshooting.md](references/troubleshooting.md) for detailed error messages, debugging steps, and performance diagnostics.

## Hardware requirements

- **Small models (7B-13B)**: 1x A10 (24GB) or A100 (40GB)
- **Medium models (30B-40B)**: 2x A100 (40GB) with tensor parallelism
- **Large models (70B+)**: 4x A100 (40GB) or 2x A100 (80GB), use AWQ/GPTQ

Supported platforms: NVIDIA (primary), AMD ROCm, Intel GPUs, TPUs

## Resources

- Official docs: https://docs.vllm.ai
- GitHub: https://github.com/vllm-project/vllm
- Paper: "Efficient Memory Management for Large Language Model Serving with PagedAttention" (SOSP 2023)
- Community: https://discuss.vllm.ai





---
Source attribution: Dicklesworthstone
Original project: https://github.com/Dicklesworthstone/pi_agent_rust/blob/f28fc46ee0db8e710d95b8b55592cd38d60f2b26/tests/ext_conformance/artifacts/templates-davila7/cli-tool/components/skills/ai-research/inference-serving-vllm/SKILL.md
License: MIT
License information: https://opensource.org/license/mit
Original upstream instructions; downloaded from Dave AI.

## Upstream license notice

MIT License (with OpenAI/Anthropic Rider)

Copyright (c) 2026 Jeffrey Emanuel

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

ADDITIONAL RIDER / RESTRICTION (OpenAI / Anthropic):

This rider is part of the "conditions" of this License. In the event of any
conflict between this rider and any other portion of this License, this rider
controls.

"Restricted Parties" means OpenAI, L.L.C.; Anthropic, PBC; any of their
respective Affiliates; and any person or entity acting directly or indirectly
on behalf of, for the benefit of, or under the direction of any of the
foregoing (including any officer, director, employee, contractor, agent,
consultant, service provider, or representative).

Notwithstanding any other provision of this License, no rights are granted to
any Restricted Party. Any purported license, sublicense, assignment, transfer,
or other permission to any Restricted Party is null and void absent the
express prior written permission of Jeffrey Emanuel.

You may not provide, disclose, distribute, sublicense, sell, lease, lend,
host, make available, or otherwise permit access to the Software or any
derivative work of the Software (as defined in applicable copyright law)
(collectively, "Derivative Works") to or for any Restricted Party.

For purposes of this rider, "use" includes, without limitation: copying,
modifying, merging, publishing, distributing, sublicensing, selling,
transferring, making available, hosting, deploying, executing, benchmarking,
testing, analyzing, indexing, or incorporating the Software or any Derivative
Works into any dataset, training corpus, evaluation harness, or pipeline for
machine learning or other automated systems.

This rider applies to the Software and all Derivative Works. As a condition of
use, you agree that this rider is a precondition to exercising any rights
under this License, and you agree that any distribution of the Software or any
Derivative Works must include this rider provision unmodified.

Any breach of this rider automatically and immediately terminates the
permissions granted by this License. Upon termination, you must immediately
cease all use and distribution of the Software and any Derivative Works and
destroy all copies under your control.

You agree that a breach of this rider would cause irreparable harm and that
Jeffrey Emanuel may seek injunctive or other equitable relief to enforce this
rider, in addition to any other remedies available at law. To the maximum
extent permitted by applicable law, the prevailing party in any action to
enforce this rider shall be entitled to recover reasonable attorneys' fees and
costs.

For purposes of this rider, "Affiliate" means any entity that directly or
indirectly controls, is controlled by, or is under common control with the
specified party. "Control" means ownership of more than 50% of the voting
securities or other ownership interest, or the power to direct management or
policies by contract or otherwise.

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.



# Dave AI resource metadata

Creator / publisher: Dicklesworthstone
Original project: https://github.com/Dicklesworthstone/pi_agent_rust/blob/f28fc46ee0db8e710d95b8b55592cd38d60f2b26/tests/ext_conformance/artifacts/templates-davila7/cli-tool/components/skills/ai-research/inference-serving-vllm/SKILL.md
Attribution basis: Source repository record

Fields of work: Development & Engineering
Roles: Backend Developer, Data Engineer, ML / AI Engineer
Specialties: API, Backend
Tools: Torch, Transformers, Vllm
Country applicability: Country not specified
Classification basis: Source catalog labels
Instruction SHA-256: 5281a24deaf3e6e10948bf087e4eff8b7d29549df12c00aad06ca1f908c7e73d