StepFun: Step 3.7 Flash

Step 3.7 Flash is a multimodal model from StepFun that accepts text, image, and video inputs with a 256,000-token context window that extends to the same length on output. It supports tool use and reasoning, giving it a footprint suited to agentic workflows, though structured output support is unconfirmed. On benchmarks, the model scores 48.0 on a blended basis across three benchmarks, with a notably stronger coding result (61.6) than its agentic score (35.5), suggesting it is more reliable for code-focused tasks than for multi-step autonomous work. At $0.20 per million input tokens and $1.15 per million output tokens, it sits in a budget-to-mid range tier, making it worth shortlisting for teams that need long-context, multimodal processing at moderate cost, particularly for coding use cases. Benchmark coverage is limited to three tests, so buyers requiring broader evidence of general capability should treat those scores as a partial picture.