Top all-time · #3937
vLLM Blog
Company blog · blog.vllm.aiaiprogramming
vLLM Blog is the company blog at blog.vllm.ai, writing about ai and programming. Since 2023, 20 different Hacker News readers have linked it 25 times, most often “VLLM: Easy, Fast, and Cheap LLM Serving with PagedAttention”. It is still writing: the last post was on 2026-10-29, and you can follow it by feed.
Too few sampled comments (6) to say how often HN recommends it.?
Why HN links it
HN commenters link vLLM Blog posts as sources on running LLMs: AMD GPU serving, Llama 3.1 405B FP8 quality, Llama 4 release evals, and vLLM routing. The links are cited as support, not praised.
Summarised by Claude from a sample of 6 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Technical articles, release announcements, model guides, and community updates from the vLLM project.
Most-linked posts?
- VLLM: Easy, Fast, and Cheap LLM Serving with PagedAttention3 people · 3 mentions · first linked 2024-01-07
- /2024/10/23/vllm-serving-amd.html2 people · 3 mentions · first linked 2024-10-30
- Llama 3.1 Support in VLLM2 people · 2 mentions · first linked 2024-07-23
- vLLM TPU: A New Unified Backend Supporting PyTorch and JAX on TPU2 people · 2 mentions · first linked 2025-10-16
- vLLM V1: A Major Upgrade to vLLM's Core Architecture2 people · 2 mentions · first linked 2025-01-27
What Hacker News says?
AMD hardware is good. The software is getting better daily. Training still sucks (mostly due to unoptimized libraries), but inference is looking pretty decent with recent advances in tuning... https://shisa.ai/blog/posts/tuning-vllm-mi300x/ https://blog.vllm.ai/2024/10/23/vllm-serving-amd.html
Comment linking /2024/10/23/vllm-serving-amd.html
on “OpenAI builds first chip with Broadcom and TSMC, scales back foundry ambition”Submitted VLLM: Easy, Fast, and Cheap LLM Serving with PagedAttention as a story · 20 points
HN discussionBasically no one uses FP32 at inference time. BF16/FP16 is typically considered unquantized, whereas FP8 is lightly quantized. That being said there's pretty minimal quality loss at FP8 compared to 16-bit typically; Llama 3.1 405b, for example, only benchmarks around ~1% worse when run at FP8: https://blog.vllm.ai/2024/07/23/llama31.html Every major inference provider other than Hyperbolic Labs runs Llama 3.1 405b a…
Comment linking /2024/07/23/llama31.html
on “Addition is all you need for energy-efficient language models”Submitted vLLM V1: A Major Upgrade to vLLM's Core Architecture as a story · 5 points
HN discussionAMD GPUs are becoming a serious contender for LLM inference. vLLM is already showing impressive performance on AMD [1], even with consumer-grade Radeon cards (even support GGUF) [2]. This could be a game-changer for folks who want to run LLMs without shelling out for expensive NVIDIA hardware. [1] https://blog.vllm.ai/2024/10/23/vllm-serving-amd.html [2] https://embeddedllm.com/blog/vllm-now-supports-running-gguf-...
Comment linking /2024/10/23/vllm-serving-amd.html
on “Making AMD GPUs competitive for LLM inference (2023)”Submitted VLLM: Easy, Fast, and Cheap LLM Serving with PagedAttention (2023) as a story · 3 points
HN discussion1x MI300x has 192GB HBM3. 1x MI325x has 256GB HBM3e. They cost less, you can fit more into a rack and you can buy/deploy at least the 300's today and 325's early next year. AMD and library software performance for AI is improving daily [0]. I'm still trying to wrap my head around how these companies think they are going to do well in this market without more memory. [0] https://blog.vllm.ai/2024/10/23/vllm-serving-a…
Comment linking /2024/10/23/vllm-serving-amd.html
on “Cerebras Trains Llama Models to Leap over GPUs”Submitted vLLM v0.6.0: 2.7x Throughput Improvement and 5x Latency Reduction as a story · 3 points
HN discussionWhile Llama 4 had a pretty bad launch (the LM Arena gaming in particular is terrible), having run my own evals on it (using the April 5 v0.8.3 vLLM release - https://blog.vllm.ai/2025/04/05/llama4.html , so before the QKNorm fix https://github.com/vllm-project/vllm/pull/16311 ) - it seemed pretty decent to me. For English, on a combination of MixEval, LiveBench, IFEval, and EvalPlus Maverick FP8 (17B/400B) was about…
Comment linking /2025/04/05/llama4.html
on “Llama 4 Smells Bad”Submitted Inside vLLM: Anatomy of a High-Throughput LLM Inference System as a story · 2 points
HN discussionAgree that routing is becoming the critical layer here. Vllm iris is really promising for this https://blog.vllm.ai/2026/01/05/vllm-sr-iris.html There's already some good work on router benchmarking which is pretty interesting
Comment linking /2026/01/05/vllm-sr-iris.html
on “The path to ubiquitous AI (17k tokens/sec)”Submitted vLLM V1: A Major Upgrade to vLLM's Core Architecture as a story · 2 points
HN discussionSubmitted VLLM: Easy, Fast, and Cheap LLM Serving with PagedAttention as a story · 2 points
HN discussionSubmitted Notes on VLLM v.s. DeepSpeed-FastGen as a story · 2 points
HN discussionSubmitted GPT-OSS Optimizations on Nvidia Blackwell: Pushing the Pareto Frontier as a story · 1 points
HN discussionSubmitted vLLM WideEP and Large-Scale Serving Toward Maturity on Blackwell (Part I) as a story · 1 points
HN discussionSubmitted DeepSeek-v3.2 on GB300: Performance Breakthrough as a story · 1 points
HN discussionSubmitted VLLM Large Scale Serving: DeepSeek 2.2k Tok/S/H200 with Wide-EP as a story · 1 points
HN discussionSubmitted Bitwise Consistent On-Policy Reinforcement Learning with VLLM and TorchTitan as a story · 1 points
HN discussionSubmitted vLLM TPU: A New Unified Backend Supporting PyTorch and JAX on TPU as a story · 1 points
HN discussionSubmitted VLLM TPU: A New Unified Back End Supporting PyTorch and Jax on TPU as a story · 1 points
HN discussionSubmitted vLLM with torch.compile: Efficient LLM inference on PyTorch as a story · 1 points
HN discussionSubmitted VLLM 2024 Retrospective and 2025 Vision as a story · 1 points
HN discussionSubmitted Installing and Developing VLLM with Ease as a story · 1 points
HN discussionSubmitted Llama 3.1 Support in VLLM as a story · 1 points
HN discussion
