1. openai gpt-5.6 luna – 15.8t
2. deepseek v4.1 flash – 11.8t
3. tencent hy4 preview – 11.6t
4. z-ai glm 5.3 flash – 11.4t
5. deepseek v4 flash – 10.6t
pricing: $0.05 to $0.834 per million input tokens.
deepseek now holds 3 of the top 5 spots. v4.1 flash launched and immediately routed 11.8t tokens – cannibalizing the older v4 flash below it.
glm 5.3 flash held #4 but its promo pricing ended sep 9 – now $0.15/m input, double what it was last week.
hy4 preview's 40% volume drop is the biggest single-week fall in the top 5.
the leaderboard isn't splitting into lanes anymore – it's rotating. every week a new model enters and reshuffles the order.
