LLM Pricing
PricingLeaderboardsToolsProvidersReleasesGuides

© 2026 LLM Pricing

About
·Contact
·Privacy
·RSS
← Model list

Step 3.7 Flash

stepfun·stepfun/step-3.7-flash·GA·Open weights
Newer StepFun flash model for faster agents, coding, and multimodal prompts

Step 3.7 Flash by stepfun is offered by 16 providers on this page. Public prices are shown for 11 of them. Its official list price is $0.185 per 1M input tokens and $1.11 per 1M output tokens. It also has 3 free ($0) channels and 2 covered by a paid subscription; free tiers usually carry rate limits, and subscription-covered access bills $0 per token only after the subscription fee.

Artificial Analysis rates it 30.9 on the Intelligence Index, with 39.6 for coding and 21.7 for agentic tasks. Against its lowest blended price of $0.416 per 1M, that is roughly 74 index points per dollar, which is the value ratio the leaderboards rank on. Median output speed is 99 tokens per second, with 2.91s to the first token. Running one task of the Artificial Analysis suite costs about $0.0913, which reflects how many tokens its reasoning consumes rather than the unit price alone.

The context window is 256,000 tokens at the reference host, but hosts report different limits, from 256,000 to 262,144, so the usable window depends on the provider you pick. It supports reasoning, tool use, and structured output. Accepted input modalities are Text, Image, and Video. The weights are open, so it can also be self-hosted or served through a gateway of your choice. Its training knowledge cuts off at 2026-03-01.

Specs & pricing

Input / output per 1M tokens
Official price·StepFun (Global)
$0.185 / $1.11
Blended $0.416 · Cache read $0.037
Lowest paid·StepFun (Global)First-party
$0.185 / $1.11
Blended $0.416 · 1.1× spread
3 more $0 channels
Context
256,000
Output limit
256,000
Knowledge cutoff
2026-03-01
Released / updated
2026-05-29 / 2026-05-29
Capabilities
✓ Reasoning✓ Tool use✓ Structured output✓ Temperature✓ Attachments
Modalities
TextImageVideo

Quality & performanceArtificial Analysis · Intelligence Index v4.1

Intelligence30.9
Value74
Coding39.6
Agentic21.7
Output speed99 tok/s
TTFT2.91 s
Cost per task$0.0913
Value formulaIQ 30.9 ÷ min blended $0.416 = 74

Quality is independently evaluated by Artificial Analysis. Speed/latency are model-level medians.

Available at 16 providers11 with public prices · 3 free · 2 subscription-covered

ProviderTierInputOutputCache readCache writeContextOutput limitStatus
Kenari
step-3-7-flash
GatewayFree——256,000256,000
UnoRouter
step-3.7-flash:free
GatewayFree——256,000256,000
NvidiaFirst-partyFree——256,00016,384
StepFun (Global)OfficialFirst-party$0.185$1.11$0.037—256,000256,000
StepFun (China)OfficialFirst-party$0.185$1.11$0.037—256,000256,000
AmbientGateway$0.19$1.14$0.03$0262,144 ⚠262,144
RequestyGateway$0.20$1.15$0.04—262,144 ⚠256,000
Hugging Face
stepfun-ai/Step-3.7-Flash
Cloud$0.20$1.15——262,144 ⚠256,000
Vercel AI GatewayCloud$0.20$1.15$0.04—256,000256,000
EmpirioLabs AI
step-3-7-flash
Gateway$0.20$1.15$0.04—256,000131,072

Sorted by blended price (input×0.75 + output×0.25) asc. The official channel always shows regardless of rank. Whether a gateway's low price is actually usable can't be verified.

Your usage cost

1StepFun (Global) · Official$74.74
2StepFun (China) · Official$74.74
3Ambient$75.80
4Requesty$78.30
5Vercel AI Gateway$78.30
6EmpirioLabs AI$78.30

3 more channels offer $0 (Kenari, UnoRouter, Nvidia); free tiers usually have rate limits and no SLA, excluded from ranking.

The cheapest paid channel is official.

Benchmark11 items

NameConditionsScoreMetricSource
SWE-Bench Pro—56.3resolve rateSource ↗
SWE-Bench Verified—76.5resolvedSource ↗
Terminal-Benchv2.159.6success rateSource ↗
Humanity's Last Examvariant: with tools47.2accuracySource ↗
BrowseComp—75.8accuracySource ↗
Toolathlon—49.5success rateSource ↗
GDPval—45.8wins or tiesSource ↗
ClawEvalv1.167.1pass^3Source ↗
Artificial Analysis Coding Index—37.1indexSource ↗
SciCode—40percent correctSource ↗
Terminal-Bench Hard—35.6success rateSource ↗

The same benchmark scores very differently across harness / dataset, so the qualifying conditions must be shown together.

Reasoning control

effort = loweffort = mediumeffort = higheffort = minimaleffort = xhigheffort = maxeffort = nonebudget_tokens

2 / 16 providers expose no reasoning control (reasoning_options: []).

Related models

DeepSeek V4 Flashcheaper alternative$0.14 / $0.28GLM-5.3-Flashcheaper alternative$0.075 / $0.25GPT-5 Nanocheaper alternative$0.05 / $0.40MiMo-V2.5cheaper alternative$0.14 / $0.28

Price historyone sample accumulated per data sync

Input listOutput listMin blended
$1.11$02026-08-052026-08-282026-08-05 · Input list $0.1852026-08-06 · Input list $0.1852026-08-07 · Input list $0.1852026-08-08 · Input list $0.1852026-08-09 · Input list $0.1852026-08-10 · Input list $0.1852026-08-11 · Input list $0.1852026-08-12 · Input list $0.1852026-08-13 · Input list $0.1852026-08-21 · Input list $0.1852026-08-28 · Input list $0.1852026-08-05 · Output list $1.112026-08-06 · Output list $1.112026-08-07 · Output list $1.112026-08-08 · Output list $1.112026-08-09 · Output list $1.112026-08-10 · Output list $1.112026-08-11 · Output list $1.112026-08-12 · Output list $1.112026-08-13 · Output list $1.112026-08-21 · Output list $1.112026-08-28 · Output list $1.112026-08-05 · Min blended $0.4162026-08-06 · Min blended $0.4162026-08-07 · Min blended $0.4162026-08-08 · Min blended $0.4162026-08-09 · Min blended $0.4162026-08-10 · Min blended $0.4162026-08-11 · Min blended $0.4162026-08-12 · Min blended $0.4162026-08-13 · Min blended $0.4162026-08-21 · Min blended $0.3942026-08-28 · Min blended $0.416
Data partly from models.dev (MIT) · StepFun (Global) official docs ↗