Step Models by StepFun
StepFunProprietary
11 models2024–2026Up to 256k ctxFrom $0.1/1M input
Last refreshed 2026-06-29. Next refresh: weekly.
Details
ResearcherStepFun
LicenseProprietary
Commercial useCommercial use: conditional
Models11
Released2024–2026
Max context256k
Capabilities
Vision3 of 11 models
Multimodal5 of 11 models
Reasoning2 of 11 models
JSON / Tool use2 of 11 models
Structured Outputs1 of 11 models
About
The Step family of large language and multimodal models from StepFun (阶跃星辰). The series spans proprietary API models and open-weight Flash releases, including Step 3.7 Flash, a 198B-parameter sparse MoE vision-language model with 256K context, Apache 2.0 weights, and selectable reasoning levels for agentic coding, tool use, image, and video workflows.
Decision facts
- Best fit
- vision and multimodal workreasoningJSON / Tool use
- Capability starting point
- Step 3.7 Flash with 256k context and reasoning, JSON / Tool use, structured outputs, and multimodal inputs
- Lowest tracked input
- Step 3.5 Flash · $0.1/1M · OpenRouter
- Closest related family
- StepAudio 2.5
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
11 in view
Step 3.7 FlashCurrent
Use when the workload needs 256k context, reasoning, and JSON / Tool use.
2026-05256k contextreasoningJSON / Tool use
Step 3.5 FlashCurrent
Use when the workload needs 256k context and reasoning.
2026-01256k contextreasoning
Step-2Current
Use when the workload needs 256k context, JSON / Tool use, and multimodal inputs.
2024-09256k contextJSON / Tool usemultimodal inputs
Step-1.5VCurrent
Use when the workload needs 128k context and multimodal inputs.
2024-06128k contextmultimodal inputs
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| Step 3.7 Flash | Use when the workload needs 256k context, reasoning, and JSON / Tool use. | 2026-05 | 256k contextreasoningJSON / Tool use | Current |
| Step 3.5 Flash | Use when the workload needs 256k context and reasoning. | 2026-01 | 256k contextreasoning | Current |
| StepFun Step-2 | Use when the workload needs 128k context. | 2025-10 | 128k context | Current |
| StepFun Step-1 | Use when the workload needs 128k context. | 2025-08 | 128k context | Current |
| Step-2 | Use when the workload needs 256k context, JSON / Tool use, and multimodal inputs. | 2024-09 | 256k contextJSON / Tool usemultimodal inputs | Current |
| Step-1V Turbo | Use when the workload needs multimodal inputs. | 2024-07 | multimodal inputs | Current |
| Step-1.5V | Use when the workload needs 128k context and multimodal inputs. | 2024-06 | 128k contextmultimodal inputs | Current |
| Step-1 | Use when the workload needs 128k context. | 2024-04 | 128k context | Current |
| Step-1V | Use when the workload needs multimodal inputs. | 2024-03 | multimodal inputs | Current |
| Step-Instruct | Use when provider availability and model metadata match the workload. | 2024-03 | — | Current |
| Step-Math | Use when provider availability and model metadata match the workload. | 2024-03 | — | Current |
Release Timeline
9 release groups2026-05
1 current
Step 3.7 Flash
Current256k contextreasoningJSON / Tool use
2026-01
1 current
Step 3.5 Flash
Current256k contextreasoning
2025-10
1 current
StepFun Step-2
Current128k context
2025-08
1 current
StepFun Step-1
Current128k context
2024-09
1 current
Step-2
Current256k contextJSON / Tool usemultimodal inputs
2024-07
1 current
Step-1V Turbo
Currentmultimodal inputs
2024-06
1 current
Step-1.5V
Current128k contextmultimodal inputs
2024-04
1 current
Step-1
Current128k context
Specifications(11 models)
| Model | Released | Context | Parameters | Vision | Multimodal | Reasoning | JSON / Tool use | Structured Outputs |
|---|---|---|---|---|---|---|---|---|
| Step 3.7 Flash | 2026-05 | 256k | 198B (11B active) | Yes | Yes | Yes | Yes | Yes |
| Step 3.5 Flash | 2026-01 | 256k | 196B (11B active) | No | No | Yes | No | No |
| StepFun Step-2 | 2025-10 | 128k | 1T (MoE)* | No | No | No | No | No |
| StepFun Step-1 | 2025-08 | 128k | — | No | No | No | No | No |
| Step-2 | 2024-09 | 256k | 1T (MoE)* | Yes | Yes | No | Yes | No |
| Step-1V Turbo | 2024-07 | — | — | No | Yes | No | No | No |
| Step-1.5V | 2024-06 | 128k | — | Yes | Yes | No | No | No |
| Step-1 | 2024-04 | 128k | — | No | No | No | No | No |
| Step-1V | 2024-03 | — | — | No | Yes | No | No | No |
| Step-Instruct | 2024-03 | — | — | No | No | No | No | No |
| Step-Math | 2024-03 | — | — | No | No | No | No | No |
Available From(3 providers)
Pricing
| Model | Provider | Input / 1M | Output / 1M | Type |
|---|---|---|---|---|
| Step 3.5 Flash | OpenRouter | $0.1 | $0.3 | Serverless |
| Step 3.7 Flash | StepFun | $0.2 | $1.15 | Serverless |
| Step 3.7 Flash | OpenRouter | $0.2 | $1.15 | Serverless |
Popular comparisons in this family
- Claude Opus 4.7 vs Step 3.5 Flash450
- Claude Sonnet 4.6 vs Step 3.5 Flash431
- GPT-5.3-Codex-Spark vs StepFun Step-2403
- Claude Sonnet 4.5 vs Step 3.5 Flash248
- Grok 4 vs Step 3.5 Flash233
- DeepSeek V3 vs StepFun Step-2213
- Claude Opus 4.6 vs Step 3.5 Flash205
- GPT-5.3-Codex-Spark vs Step 3.5 Flash175
- Kimi K2.5 vs Step 3.5 Flash169
- DeepSeek V3.1 vs Step 3.5 Flash147
Comparisons
- Step 3.7 Flash vs Step 3.5 Flash
- Step 3.7 Flash vs Gemini 2.5 Flash
- Step 3.7 Flash vs Kimi K2.6
- Step 3.7 Flash vs Qwen3-235B-A22B
- Step 3.7 Flash vs DeepSeek V3
- Step 3.7 Flash vs GPT-4o-mini
- Step 3.7 Flash vs MiniMax M2.7
- Step 3.7 Flash vs Claude 3.5 Haiku
Models(11)
Step 3.7 Flash
2026-05256k198B (11B active)3 providers
MultimodalReasoning
Step 3.5 Flash
2026-01256k196B (11B active)1 provider
Reasoning
StepFun Step-2
2025-10128k1T (MoE)*
StepFun Step-1
2025-08128k
Step-2
2024-09256k1T (MoE)*1 provider
Multimodal
Step-1V Turbo
2024-07
Multimodal
Step-1.5V
2024-06128k1 provider
Multimodal
Step-1
2024-04128k1 provider
Step-1V
2024-03
Multimodal
Step-Instruct
2024-03
Step-Math
2024-03