Step 3.5 Flash
Intelligence
37.8
#99/521
Coding
31.6
#97/427
Math
—
Speed
79 tok/s
TTFT: 3.07s
Pricing
$0.10 / $0.30
per 1M tokens (in/out)
Step-3.5-Flash is StepFun’s fast, cost-effective text model optimized for quick inference. Built on their Step3 architecture, it offers strong performance across text tasks with low latency and efficient token usage, ideal for production workloads requiring speed and cost efficiency.
Step 3.5 Flash is StepFun’s model designed for efficient token processing. It operates at a speed of 78.582 tokens per second and is priced at $0.1 per million input tokens and $0.3 per million output tokens, catering to users who require fast and cost-effective solutions.
When to Use Step 3.5 Flash
✓ Best For
- Natural language processing tasks
- Real-time data analysis
- Cost-sensitive applications
✗ Not Ideal For
- Complex mathematical computations
- High-level coding tasks
How Step 3.5 Flash Compares
Intelligence Index · Higher is better
OpenAIxAIStepFunInclusionAI
Benchmark Profile
Coding Index
AlibabaOpenAIStepFunDeepSeekTencent
Output Speed · tok/s
OpenAIStepFunGoogleAlibaba
Intelligence · Coding · Math
Intelligence
Coding
Math
All Benchmark Scores (18)
| Benchmark | Score | |
|---|---|---|
| Intelligence Index | 37.8 | |
| Coding Index | 31.6 | |
| GPQA | 831% | |
| HLE | 191% | |
| SWE-Bench Verified | 74.4% | |
| SciCode | 38.5% | |
| IFBench | 66.5% | |
| LCR | 54.3% | |
| TerminalBench Hard | 32.6% | |
| Tau2 | 87.4% | |
| AIME 2025 | 97.3% | |
| BrowseComp | 69% | |
| IMO-AnswerBench | 85.4% | |
| LiveCodeBench v6 | 86.4% | |
| Tau-bench | 88.2% | |
| Terminal-Bench 2.0 | 51% | |
| Arena Chat | 10.5 | |
| Arena Coding | 7.3 |
Data: Artificial Analysis · Updated: March 26, 2026
Frequently Asked Questions (15)
When was Step 3.5 Flash released?
Step 3.5 Flash was released on February 2, 2026.
Who created Step 3.5 Flash?
Step 3.5 Flash was created by StepFun.
How intelligent is Step 3.5 Flash?
Step 3.5 Flash scores 38 on the Artificial Analysis Intelligence Index, placing it well above average among other open weight models of similar size (median: 27).
How fast is Step 3.5 Flash?
Step 3.5 Flash generates output at 92.6 tokens per second (based on StepFun's API), which is well above average compared to other open weight models of similar size (median: 55.0 t/s).
What is the latency of Step 3.5 Flash?
Step 3.5 Flash has a time to first token (TTFT) of 2.86s (based on StepFun's API), which is at the higher end compared to other open weight models of similar size (median: 2.24s).
How much does Step 3.5 Flash cost?
Step 3.5 Flash costs $0.10 per 1M input tokens (very competitive, median: $0.60) and $0.30 per 1M output tokens (very competitive, median: $2.20), based on StepFun's API.
What is Step 3.5 Flash API pricing?
Step 3.5 Flash costs $0.10 per 1M input tokens and $0.30 per 1M output tokens (based on StepFun's API). For a blended rate (3:1 input to output ratio), this is $0.15 per 1M tokens. Pricing may vary by provider.
How verbose is Step 3.5 Flash?
When evaluated on the Intelligence Index, Step 3.5 Flash generated 200M output tokens, which is at the higher end compared to other open weight models of similar size (median: 17M).
Is Step 3.5 Flash a reasoning model?
Yes, Step 3.5 Flash is a reasoning model. It uses extended thinking or chain-of-thought reasoning to work through complex problems before providing an answer.
What input modalities does Step 3.5 Flash support?
Step 3.5 Flash supports text input.
What output modalities does Step 3.5 Flash support?
Step 3.5 Flash supports text output.
Can Step 3.5 Flash process images?
No, Step 3.5 Flash does not support image input. It can only process text.
Is Step 3.5 Flash multimodal?
No, Step 3.5 Flash is not multimodal. It only supports text input.
What is the context window of Step 3.5 Flash?
Step 3.5 Flash has a context window of 260k tokens. This determines how much text and conversation history the model can process in a single request.
Is Step 3.5 Flash open source?
Yes, Step 3.5 Flash is open weights. The model weights are publicly available and can be downloaded for self-hosting.
