Ling 3.1 Flash
InclusionAI · 2026-09-30
InclusionAI's hybrid reasoning language model for coding, analysis and tool-using agents, available through the inclusionai/ling-3.1-flash route on Vercel AI Gateway and OpenRouter. It first became publicly available when Vercel AI Gateway announced it on September 30, 2026; inclusionAI later published MIT-licensed FP8 weights on Hugging Face in mid-October, and that card links the same OpenRouter route. The card describes rounded figures of 560B total and 25B activated parameters, with no exact unquantized parameter count published; the released checkpoint supports 262,144 tokens natively and 1,048,576 with documented fourfold YaRN extrapolation, while gateway routes may cap context at 262,144 tokens.