Step 3.7 Flash
Step 3.7 Flash is StepFun's latest high-efficiency multimodal Mixture-of-Experts model. It pairs a 196B-parameter language backbone with a vision encoder for native image and video understanding, activating roughly 11B parameters per token. The model supports a 256K context window and exposes selectable reasoning levels (high/medium/low), letting callers trade off speed, cost, and depth of reasoning. Designed for coding, agentic workflows, structured outputs, and long-context productivity tasks.
- Input / output price
- $0.16 / $0.92 per 1M
- Context
- 262K tokens
- Released
- May 29, 2026
Paste into Claude Code, Cursor, or Codex
Help me set up the Respan gateway: fetch and follow https:/
Providers
5 providers, cheapest first, OpenRouter last, priced per 1M tokens after gateway discounts.
| Provider | Input | Output |
|---|---|---|
Concentrate concentrate/step-3-7-flash | $0.16/M | $0.92/M |
EmpirioLabs empiriolabs/step-3.7-flash | $0.20/M | $1.15/M |
Novita novita/step-3.7-flash | $0.20/M | $1.15/M |
StepFun stepfun/step-3.7-flash | $0.20/M | $1.15/M |
OpenRouter openrouter/step-3.7-flash | $0.20/M | $1.15/M |
Is a provider down? OpenRouter status
Pricing
Averaged across 5 providers, in USD per 1M tokens after gateway discounts.
- Average input
- $0.19 per 1M
- Average output
- $1.10 per 1M
- Average cache read
- $0.04 per 1M
Cost by volume
Total cost from 0 to 1T tokens on the cheapest and most expensive provider.
Throughput and latency
30-day averages from requests through the Respan gateway. Under each figure, every model on the gateway by its average, this one in blue.
Output speed
Higher is better7.2tokens/sacross providers
Response latency
Lower is better1.5sacross providers
Benchmarks
19 published scores from 2 sources, each linked to where it was published.
Published scores
| Benchmark | Score | Source | Published |
|---|---|---|---|
| SWE-bench Verifiedstatic.stepfun.com | 76.5% | static.stepfun.com | May 29, 2026 |
| SWE-bench Multilingualstatic.stepfun.com | 72.4% | static.stepfun.com | May 29, 2026 |
| Terminal-Bench 2.1static.stepfun.com | 59.6% | static.stepfun.com | May 29, 2026 |
| SWE-bench Prostatic.stepfun.com | 56.3% | static.stepfun.com | May 29, 2026 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| DeepSearchQA F1static.stepfun.com | 92.8% | static.stepfun.com | May 29, 2026 |
| DeepSearchQA Accuracystatic.stepfun.com | 81.7% | static.stepfun.com | May 29, 2026 |
| BrowseCompstatic.stepfun.com | 75.8% | static.stepfun.com | May 29, 2026 |
| ResearchRubricsstatic.stepfun.com | 71.7% | static.stepfun.com | May 29, 2026 |
| ClawEval v1.1static.stepfun.com | 67.1% | static.stepfun.com | May 29, 2026 |
| Toolathlonstatic.stepfun.com | 49.5% | static.stepfun.com | May 29, 2026 |
| GDPval-Stirrupstatic.stepfun.com | 1416 Elo | static.stepfun.com | May 29, 2026 |
| GDPval-Stirrup Intelligence Indexstatic.stepfun.com | 45.8 | static.stepfun.com | May 29, 2026 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| AA-LCRstatic.stepfun.com | 63.9% | static.stepfun.com | May 29, 2026 |
| HLE (with tools, text-only)static.stepfun.com | 49.7% | static.stepfun.com | May 29, 2026 |
| HLE (with tools)static.stepfun.com | 47.2% | static.stepfun.com | May 29, 2026 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| V* (with Python)static.stepfun.com | 95.3% | static.stepfun.com | May 29, 2026 |
| SimpleVQA (with tools)static.stepfun.com | 79.2% | static.stepfun.com | May 29, 2026 |
| WorldVQA (with visual search)static.stepfun.com | 58.1% | static.stepfun.com | May 29, 2026 |
Apps
Use Step 3.7 Flash in the tools you already run. Each guide points the app at the Respan gateway.
Setup guides for more apps are in the integration docs.
Step 3.7 Flash FAQ
Short answers, built from the data on this page.
Step 3.7 Flash is a model from StepFun, released on May 29, 2026. Step 3.7 Flash is StepFun's latest high-efficiency multimodal Mixture-of-Experts model. It pairs a 196B-parameter language backbone with a vision encoder for native image and video understanding, activating roughly 11B parameters per token. The model supports a 256K context window and exposes selectable reasoning levels (high/medium/low), letting callers trade off speed, cost, and depth of reasoning. Designed for coding, agentic workflows, structured outputs, and long-context productivity tasks.
On the Respan gateway, Step 3.7 Flash costs from $0.16 per 1M input tokens and $0.92 per 1M output tokens on Concentrate, after gateway discounts. Averaged across its 5 providers, it costs $0.19 per 1M input tokens and $1.10 per 1M output tokens.
Step 3.7 Flash is available through 5 providers: Concentrate, EmpirioLabs, Novita, StepFun, and OpenRouter.
Step 3.7 Flash accepts up to 262K tokens of context on its largest provider route.
Point any OpenAI-compatible SDK at https://gateway.respan.ai/v1 with your Respan API key and set the model to "stepfun/step-3.7-flash". To set it up with a coding agent, copy the agent setup prompt at the top of this page.
Step 3.7 Flash has 19 published benchmark scores, including SWE-bench Verified 76.5%, DeepSearchQA F1 92.8%, and AA-LCR 63.9%. Each score links to where it was published; the data is collected by Respan.
Certified where it counts.
Built to meet the security and privacy standards that enterprise and healthcare teams require.
ISO 27001
The internationally recognized standard for information security management.
SOC 2
Secure, compliant management of your data across all of our systems.
GDPR
Operated under GDPR, the world's strictest standard for data privacy.
HIPAA
HIPAA compliant, with a BAA available for healthcare teams.