Kimi K3 runs on AMD MI350X with SGLang and hits 327 tok/s across four requests
burny_tech · x · 2026-07-28
Kimi K3 is shown running on AMD MI350X via SGLang, with the author reporting that it worked out of the box.
The attached benchmark screenshot suggests solid serving performance:
- Single-stream throughput ranges from about 105.6 tok/s on short prompts to 143.4 tok/s on reasoning prompts.
- In a 4-way parallel test, aggregate throughput reaches 327 tok/s across four requests.
The post is mainly a deployment note for AMD-based inference rather than a model announcement, and it implicitly signals growing compatibility between Kimi, SGLang, and AMD hardware.
Related event: SGLang Day-0 Support for Kimi K3 Boosts Throughput to 423 tok/s(8 posts)→
More from Infra
- Engram's random reads don't suit SSDs; CPU-memory over NVLink could serve all 72 GPUs — bookwormengr · 2026-09-11
- 80% of the DIY LLM inference hype posters have already quit — it's brutally hard systems work — abhijithneil · 2026-09-11
- Hugging Face's Ultra Scale Playbook: a free book on training LLMs on GPU clusters — mdancho84 · 2026-09-11
- Is inference latency becoming the biggest bottleneck for production AI agents? — Euphoric_Sea632 · 2026-09-11
- LLM Serving Metrics Thread: Why TPOT and Uptime Make or Break User Experience — abhijithneil · 2026-09-11
- PlanetScale launches sharded Postgres: 768 servers acting as one, 1PB scale — dhruv2038 · 2026-09-11