Icon for Xinference

Xinference

Open Source Free Trial

Open-source inference engine for 300+ open models, plus a hosted API and managed GPUs served from Australia

Xinference is an Apache-2.0 inference engine for serving open models behind an OpenAI-compatible API, developed in the open at xorbitsai/inference (about 9,600 GitHub stars). The Community edition runs on NVIDIA GPUs in a single cluster.

Xinference Holdings, a Singapore company, also sells a hosted Model API and dedicated H100, H200, B200 and A100 instances served from Australia. An Enterprise edition adds AMD, Intel and CPU support plus multi-cluster scaling.

The engine is free. The Model API is billed per token, from $0.30 per 1M tokens; dedicated GPUs and Enterprise pricing go through sales.

Pricing: Free (open source); Model API from $0.30 per 1M tokens; dedicated and Enterprise via sales

Hosting Cloud + Self-hosted
Pricing Freemium, Free (open source)
HQ ๐Ÿ‡ธ๐Ÿ‡ฌ Singapore
License APACHE-2.0
Screenshot of Xinference webpage

Xinference prices by model

Per 1M tokens, read off Xinference's own pricing page on the date shown.

Model Input / 1M Cached / 1M Output / 1M Checked Notes
DeepSeek V4.1 Flash $0.30 โ€“ $1.20 9 Nov 2026 Hosted in Australia
GLM 5.3 $1.40 โ€“ $4.40 9 Nov 2026 Hosted in Australia

Work on Xinference? Feature it at the top of Inference APIs.

Is your product missing?

Add it here →