← Model list
Step 3.5 Flash 2603
stepfun·stepfun/step-3.5-flash-2603·GA·Open weights
StepFun flash model for efficient multimodal reasoning, coding, and tool use
At a glance
Official price$0.10 / $0.30 per 1M
Cache read $0.02 · StepFun (Global)
Lowest paid$0.10 / $0.30
NanoGPT Gateway · 1× spread
Context256,000
Output limit256,000
Capabilities
✓ Reasoning✓ Tool useStructured output✓ TemperatureAttachments
⚠ Providers report capability flags inconsistently
Modalities
Text
Knowledge cutoff2025-01
Released / updated2026-04-02 / 2026-04-02
Quality & performanceArtificial Analysis · Intelligence Index v4.1
Intelligence26.5
Value177
Coding—
Agentic—
Output speed214 tok/s
TTFT1.14 s
Value formulaIQ 26.5 ÷ min blended $0.150 = 177
Quality is independently evaluated by Artificial Analysis. Speed/latency are model-level medians.
Available at 6 providers4 with public prices · 2 subscription-covered
| Provider | Tier | Input | Output | Cache read | Cache write | Context | Output limit | Status |
|---|---|---|---|---|---|---|---|---|
| NanoGPT | Gateway | $0.10 | $0.30 | $0.05 | — | 256,000 | 256,000 | |
| StepFun (Global)Official | First-party | $0.10 | $0.30 | $0.02 | — | 256,000 | 256,000 | |
| StepFun (China)Official | First-party | $0.10 | $0.30 | $0.02 | — | 256,000 | 256,000 | |
| EmpirioLabs AI step-3-5-flash-2603 | Gateway | $0.10 | $0.30 | $0.02 | — | 256,000 | 131,072 | |
| StepFun Step Plan (China) | Gateway | — | — | — | — | 256,000 | 256,000 | hostTable.undisclosed |
| StepFun Step Plan (Global) | Gateway | — | — | — | — | 256,000 | 256,000 | hostTable.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.
Your usage cost
1StepFun (Global) · Official$25.40
2StepFun (China) · Official$25.40
3EmpirioLabs AI$25.40
4NanoGPT$29.00
The cheapest paid channel is official.
Benchmark3 items
| Name | Conditions | Score | Metric | Source |
|---|---|---|---|---|
| Artificial Analysis Coding Index | — | 34.6 | index | Source ↗ |
| SciCode | — | 38.5 | percent correct | Source ↗ |
| Terminal-Bench Hard | — | 32.6 | success rate | Source ↗ |
The same benchmark scores very differently across harness / dataset, so the qualifying conditions must be shown together.
Reasoning control
effort = loweffort = mediumeffort = high
Related models
GLM-4.7-Flashcheaper alternative$0 / $0Hy3cheaper alternative$0 / $0Nemotron 3 Nano 30B A3Bcheaper alternative$0 / $0Qwen3.7 Flashcheaper alternative$0.03 / $0.118
Price historyone sample accumulated per data sync
Input listOutput listMin blended