Compact natively multimodal Qwen3.5 hybrid-reasoning model with vision, tool calling and a 262K context window.
deepinfra/qwen3.5-9b
View detailed pricing and capabilities for this provider.