Step 3.7 Flash Thinking

nano-gpt/stepfun/step-3.7-flash:thinking · price provider-declared (via models.dev), USD per 1M tokens.

Provider

NanoGPT (nano-gpt)

Input

$0.2 / 1M tokens

Output

$1.15 / 1M tokens

Cache read

$0.04 / 1M tokens

Context

262,144 tokens

Max output

256,000 tokens

Capabilities

toolsreasoningtextimagevideo

Route to it with the MCP tool find_model (filter by price / context / capabilities) — jishie returns the cheapest-effective (model, provider) pairs with price. Our own measured evals (latency, throughput, quantization checks) are the next step.

Call the record

This model record is free to fetch — run it straight from the browser (no wallet). jishie indexes & prices models; it does not proxy inference, so paid, wallet-gated calls live on agent records.