LLM Pricing
ModelsLeaderboardsToolsProvidersReleasesArticles

© 2026 LLM Pricing

About
·Contact
·Privacy
·RSS
← Model list

Step 3.5 Flash 2603

stepfun·stepfun/step-3.5-flash-2603·GA·Open weights·step series
StepFun flash model for efficient multimodal reasoning, coding, and tool use

Specs & pricing

Input / output per 1M tokens
Official price·StepFun (China)
$0.10 / $0.31
Blended $0.16 · Cache read $0.021
Lowest paid·StepFun (Global)First-party
$0.10 / $0.30
Blended $0.15 · 1× spread
Context
256,000
Output limit
256,000
Knowledge cutoff
2025-01
Released / updated
2026-04-02 / 2026-04-02
Capabilities
✓ Reasoning✓ Tool use✓ Structured output✓ TemperatureAttachments
Modalities
Text
Weights
Hugging Face

Step 3.5 Flash 2603 by stepfun is offered by 5 providers on this page. Public prices are shown for 3 of them. Its official list price is $0.105 per 1M input tokens and $0.314 per 1M output tokens. It also has 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.

Median output speed is 148 tokens per second, with 3.27s to the first token.

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. 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 2025-01.

Quality & performanceArtificial Analysis · Intelligence Index v5

Intelligence—
Value—
Output speed148 tok/s
TTFT3.27 s

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

Available at 5 providers3 with public prices · 2 subscription-covered

  • ProviderInput / Output
  • StepFun (Global)OfficialFirst-party
    $0.10 / $0.30
    Cache read $0.02Context 256,000Output limit 256,000
  • EmpirioLabs AIGateway
    step-3-5-flash-2603
    $0.10 / $0.30
    Cache read $0.02Context 262,144 ⚠Output limit 131,072
  • StepFun (China)OfficialFirst-party
    $0.10 / $0.31
    Cache read $0.021Context 256,000Output limit 256,000
  • StepFun Step Plan (China)Gateway
    — / —
    Context 256,000Output limit 256,000hostTable.undisclosed
  • StepFun Step Plan (Global)Gateway
    — / —
    Context 256,000Output limit 256,000hostTable.undisclosed
ProviderTierInputOutputCache readCache writeContextOutput limitStatus
StepFun (Global)OfficialFirst-party$0.10$0.30$0.02—256,000256,000
EmpirioLabs AI
step-3-5-flash-2603
Gateway$0.10$0.30$0.02—262,144 ⚠131,072
StepFun (China)OfficialFirst-party$0.10$0.31$0.021—256,000256,000
StepFun Step Plan (China)Gateway————256,000256,000hostTable.undisclosed
StepFun Step Plan (Global)Gateway————256,000256,000hostTable.undisclosed

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.

Reasoning control

effort = loweffort = high

Interleaved thinking (reasoning between tool calls) is declared by 4 of 5 providers.

Your usage cost

Common scenarios
1StepFun (Global) · Official$25.40
2EmpirioLabs AI$25.40
3StepFun (China) · Official$26.57
The cheapest paid channel is official.

Price historyone sample accumulated per data sync

Input listOutput listMin blended
$0.31$02026-08-052026-10-21
2026-10-21Input list $0.10Output list $0.31Min blended $0.15

Benchmark3 items

NameConditionsScoreMetricSource
Artificial Analysis Coding Index—34.6indexSource ↗
SciCode—38.5percent correctSource ↗
Terminal-Bench Hard—32.6success rateSource ↗

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

Related models

StepAudio 2.5 ASRsame series—StepAudio 2.5 TTSsame series—Step TTS 2same series—Gemma 4 31B ITcheaper alternative$0 / $0Nemotron 3 Ultra 550B A55Bcheaper alternative$0 / $0GLM-4.7-Flashcheaper alternative$0 / $0Hy3cheaper alternative$0 / $0Step 3.5 Flashsame lab$0.10 / $0.31Step 3.7 Flashsame lab$0.20 / $1.21
Data partly from models.dev (MIT) · StepFun (China) official docs ↗