Qwen 3.8 Flash and GLM 5.3 Flash land on OpenRouter

What shipped? OpenRouter added two new flash-class models today: Qwen 3.8 Flash (Alibaba, 27B parameters) and GLM 5.3 Flash (Z.ai). Both were also added to Vercel AI Gateway in tandem.

What changed? Qwen 3.8 Flash is a 27B reasoning-capable model at $0.35/M input tokens and $2.75/M output, with 65K max completion tokens and support for structured outputs, tools, and image inputs. GLM 5.3 Flash is already seeing heavy usage — 4.8M requests on OpenRouter today alone — and is similarly priced in the sub-$1/M flash bracket. Both models support reasoning, tool calling, and multipart input.

Why a builder cares. Flash-class models now span three major families (Qwen, GLM, and the existing DeepSeek V4 Flash), giving builders more price-competitive options for latency-sensitive, reasoning-heavy workloads without leaving the OpenAI-compatible API surface.