ChineseModels.ai
Research

Field notes on Chinese AI models

Comparisons you can reproduce, deployment guides that survive production, licence breakdowns for the people who have to sign off, and a weekly radar of what actually shipped.

6 posts4 pillarsNo sponsored rankings
AllComparisonsDeploymentComplianceRadar

Open-weight setups, quantisation, vLLM and SGLang configs that survive production.

DeploymentOct 11, 2026·2 min read

How to deploy Qwen locally: Ollama for speed, vLLM for throughput

Two paths to a working local Qwen endpoint — a five-minute Ollama setup for a single developer, and a vLLM server with prefix caching for a team.

QwenOllamavLLMRead
DeploymentOct 8, 2026·2 min read

Self-hosting Kimi on consumer GPUs: an honest reality check

Kimi's appeal is extreme long context. That is also the thing that makes local hosting expensive. Here is the memory math before you buy hardware.

KimiMoonshotlong contextRead

Everything, newest first

Oct 20, 2026
China AI model radar: what's actually live this week
Our first weekly radar — what the platform is serving, how the families are positioned on price, and what we still need to measure before publishing numbers.
Radar2 min
Oct 18, 2026
GLM-5 vs DeepSeek V4: which one actually finishes your coding tasks?
A practical teardown of the two models developers keep shortlisting for agentic coding — where they differ, what to measure, and a 40-minute harness to decide on your own repo.
Comparisons3 min
Oct 14, 2026
DeepSeek vs Qwen for local coding: VRAM, licences and what actually runs
Both families ship open weights. The decision is less about benchmarks and more about which size class fits your GPU, your licence review and your latency budget.
Comparisons3 min
Oct 11, 2026
How to deploy Qwen locally: Ollama for speed, vLLM for throughput
Two paths to a working local Qwen endpoint — a five-minute Ollama setup for a single developer, and a vLLM server with prefix caching for a team.
Deployment2 min
Oct 8, 2026
Self-hosting Kimi on consumer GPUs: an honest reality check
Kimi's appeal is extreme long context. That is also the thing that makes local hosting expensive. Here is the memory math before you buy hardware.
Deployment2 min
Oct 5, 2026
Apache 2.0 vs custom terms: reading Chinese open-weight licences
What to check before shipping a Chinese open-weight model in a commercial product — and why the family name is a worse guide than the individual model card.
Compliance3 min