Step 3.7 Flash is a StepFun model with 262K context, $0.20 in / $1.15 out per million tokens, last verified October 25, 2026. Open weights: yes. Public-board scores are linked to the publisher; vendor-only figures stay in the watch-out, not on the board.
Step 3.7 Flash is StepFun's Apache-declared open-weight multimodal MoE, with a hosted text/image/video endpoint and selectable reasoning effort.
Recorded profile verification October 25, 2026·Individual provider fields may change
AI model specification details
Specification
Step 3.7 Flash
Provider
StepFun
Tier
Budget
Context window
262K
Max output
Not verifiedUnverified
Input / 1M tokens
$0.20
Output / 1M tokens
$1.15
Weights
Open
Parameters
198B total (196B language backbone + 1.8B vision encoder) / approximately 11B active
Pricing tiers: StepFun hosted USD/MTok input/output $0.20/$1.15; cache-hit input $0.04. The Apache-declared native checkpoint has 262,144 positions while the hosted catalogue advertises 256K. Prompt caching activates above 256 tokens, uses 256-token units and LRU eviction; first writes use cache-miss input pricing. Internet search is $0.006/call, image search $0.02/call and file storage $0.08/GB/day, separately from token rates. Hosted tariffs do not price self-hosting.
Best for
High-volume multimodal
Cost-sensitive API
Chinese market
Watch out
Hosted reasoning accepts low/medium/high via reasoning_effort (Chat) or output_config.effort (Messages); medium is a documented recommendation, not a verified omitted-parameter default. An exact hosted maximum output remains unverified in the checked documents. The checkpoint card declares Apache 2.0. Up-to-400-TPS and launch benchmark results are vendor claims dependent on settings and infrastructure. Public PAYG account limits depend on cumulative cash top-ups and identity verification; subscription Step Plan Credits are separate. Cache hits are not guaranteed and do not reduce the logical input-token count.
Source receipts
Catalog figures for this model were checked against the following sources.
Answered from the verified figures on this page rather than general guidance.
How much does Step 3.7 Flash cost per million tokens?
Step 3.7 Flash lists base rates of $0.20 per million input tokens and $1.15 per million output tokens. Estimate both input and output token volume and any additional charges for your workload. StepFun hosted USD/MTok input/output $0.20/$1.15; cache-hit input $0.04. The Apache-declared native checkpoint has 262,144 positions while the hosted catalogue advertises 256K. Prompt caching activates above 256 tokens, uses 256-token units and LRU eviction; first writes use cache-miss input pricing. Internet search is $0.006/call, image search $0.02/call and file storage $0.08/GB/day, separately from token rates. Hosted tariffs do not price self-hosting.
What is Step 3.7 Flash's context window?
Step 3.7 Flash accepts about 262K tokens of context. That only matters if you routinely send very long documents, large codebases, or multi-turn histories that approach that limit.
What is Step 3.7 Flash best for?
Step 3.7 Flash is a budget tier from StepFun. It suits high-volume multimodal, cost-sensitive api, chinese market. Hosted reasoning accepts low/medium/high via reasoning_effort (Chat) or output_config.effort (Messages); medium is a documented recommendation, not a verified omitted-parameter default. An exact hosted maximum output remains unverified in the checked documents. The checkpoint card declares Apache 2.0. Up-to-400-TPS and launch benchmark results are vendor claims dependent on settings and infrastructure. Public PAYG account limits depend on cumulative cash top-ups and identity verification; subscription Step Plan Credits are separate. Cache hits are not guaranteed and do not reduce the logical input-token count.
Can I self-host Step 3.7 Flash?
Step 3.7 Flash publishes open weights, but self-hosting depends on the licence, hardware footprint, quantisation quality, and serving stack. A hosted API is often cheaper until you have measured throughput and concurrency on your own hardware.
