Chinese open-weight models now account for more than 60% of OpenRouter usage — a 60× jump in 18 months

Air Street's State of AI May 2026 report shows Chinese open-weight models — DeepSeek, Qwen, Kimi, GLM — went from roughly 1% of OpenRouter usage in mid-2024 to more than 60% in May 2026. The shift tracks a 5–20× price-per-token gap to closed flagships and a near-elimination of the capability gap on most evaluation suites.

The 60% share is the under-priced data point in the May 2026 frontier-model debate. Most policy conversations still frame Chinese models as a research-equivalence story ("closing the gap") rather than a production-deployment story ("already dominant where pricing matters"). The OpenRouter aggregate captures the production reality: developers choosing daily-usage models on cost-per-quality, not lab affiliation.

For US-side export-control posture, the data complicates the narrative. If the goal of GPU export restrictions was to slow Chinese frontier capability, the OpenRouter numbers suggest the restrictions slowed where chips ship, not where models get used. See our analysis →.

Air Street Press — State of AI May 2026 → · Codersera — best open-source LLM May 2026 →