All Models

Qwen 3.8 Flash Next

qwenReasoningTool CallingAttachmentsOpen WeightsStructured Output

Open-weight experimental preview of the Qwen4 architecture: hybrid-attention MoE (125B total, 6B active) with vision encoder for coding, agent tasks, and image and video understanding

Providers5
ReleasedAug 27, 2026
Input Modalitiestext, image, video
Output Modalitiestext
Tarsk Usecoding

Benchmarks

55.8
Intelligence Index
73.1
Coding Index

Available Providers (5)

ProviderModel IDInput CostOutput CostContextMax OutputDocs
Vercel AI Gatewayalibaba/qwen3.8-flash-next$0.12/MTok$0.40/MTok1.0M1.0M
AMDQwen3.8-Flash-Next$0.15/MTok$0.47/MTok262.1K131.1K
Requestyqwen3.8-flash-next$0.20/MTok$0.50/MTok262.1K262.1K
Requestyqwen3.8-flash-next@eu$0.20/MTok$0.50/MTok262.1K262.1K
Cortecsqwen3.8-flash-next$0.20/MTok$0.50/MTok262.1K64K

Capabilities

Reasoning
Tool Calling
Attachments
Open Weights
Structured Output