Step 3.7 Flash

Tool callingStructured outputReasoningImage inputVideo input

Hugging Face Website

Newer StepFun flash model for faster agents, coding, and multimodal prompts

Context: 262K
Released: 4mo ago
Parameters: —
Input / 1M tokens: $0.20
Output / 1M tokens: $1.15
Cached input / 1M tokens: $0.04

Sources:models.devOpenRouter

Benchmarks

DeepSWE is the primary coding benchmark where an exact model evaluation exists. Other sourced evaluations appear below; missing scores are not zero.

Not evaluated in DeepSWE 1.1 as of 2026-09-22. Results are only shown for exact model versions in the official dataset. Official leaderboard.

No benchmark results have been matched to this model yet.

Specifications

Developer: Stepfun
Model family: Step
Type: Language model
Predecessor: —
Successor: —
Released: May 29, 2026
Model ID: stepfun/step-3.7-flash
Openness: Open weights
License: —
Parameters: —
Input modalities: text, image, video
Output modalities: text
Modality: text+image+video->text
Tokenizer: Other
Max output tokens: 230K
Knowledge cutoff: 2026-03-01
Hugging Face: stepfun-ai/Step-3.7-Flash
Source freshness: Metadata synced Oct 8, 2026, 10:17 PM UTC

Additional source evidence

Source model: stepfun/step-3.7-flash
Last updated: 2026-05-29
Knowledge cutoff: 2026-03-01
Model context: 256K
Maximum input: 256K
Model output limit: 256K
Open weights: Yes
File attachments: Yes
Reasoning: Yes
Tool calling: Yes
Temperature control: Yes

Downloadable weights

Hugging Face

Published benchmarks

Benchmark	Score	Evaluation details	Evidence
SWE-Bench Pro	56.3 · resolve rate	2026-05-29	Source
SWE-Bench Verified	76.5 · resolved	2026-05-29	Source
Terminal-Bench	59.6 · success rate	2.1 · 2026-05-29	Source
Humanity's Last Exam	47.2 · accuracy	with tools · 2026-05-29	Source
BrowseComp	75.8 · accuracy	2026-05-29	Source
Toolathlon	49.5 · success rate	2026-05-29	Source
GDPval	45.8 · wins or ties	2026-05-29	Source
ClawEval	67.1 · pass^3	1.1 · 2026-05-29	Source
Artificial Analysis Coding Index	37.1 · index	2026-06-15	Source
SciCode	40 · percent correct	2026-06-15	Source
Terminal-Bench Hard	35.6 · success rate	2026-06-15	Source

Complete source record (JSON)

{
  "id": "stepfun/step-3.7-flash",
  "name": "Step 3.7 Flash",
  "description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts",
  "last_updated": "2026-05-29",
  "knowledge": "2026-03-01",
  "attachment": true,
  "reasoning": true,
  "tool_call": true,
  "temperature": true,
  "modalities": {
    "input": [
      "text",
      "image",
      "video"
    ],
    "output": [
      "text"
    ]
  },
  "limit": {
    "context": 256000,
    "input": 256000,
    "output": 256000
  },
  "benchmarks": [
    {
      "name": "SWE-Bench Pro",
      "score": 56.3,
      "source": "https://static.stepfun.com/blog/step-3.7-flash/",
      "metric": "resolve rate",
      "date": "2026-05-29"
    },
    {
      "name": "SWE-Bench Verified",
      "score": 76.5,
      "source": "https://static.stepfun.com/blog/step-3.7-flash/",
      "metric": "resolved",
      "date": "2026-05-29"
    },
    {
      "name": "Terminal-Bench",
      "score": 59.6,
      "source": "https://static.stepfun.com/blog/step-3.7-flash/",
      "metric": "success rate",
      "version": "2.1",
      "date": "2026-05-29"
    },
    {
      "name": "Humanity's Last Exam",
      "score": 47.2,
      "source": "https://static.stepfun.com/blog/step-3.7-flash/",
      "metric": "accuracy",
      "variant": "with tools",
      "date": "2026-05-29"
    },
    {
      "name": "BrowseComp",
      "score": 75.8,
      "source": "https://static.stepfun.com/blog/step-3.7-flash/",
      "metric": "accuracy",
      "date": "2026-05-29"
    },
    {
      "name": "Toolathlon",
      "score": 49.5,
      "source": "https://static.stepfun.com/blog/step-3.7-flash/",
      "metric": "success rate",
      "date": "2026-05-29"
    },
    {
      "name": "GDPval",
      "score": 45.8,
      "source": "https://static.stepfun.com/blog/step-3.7-flash/",
      "metric": "wins or ties",
      "date": "2026-05-29"
    },
    {
      "name": "ClawEval",
      "score": 67.1,
      "source": "https://static.stepfun.com/blog/step-3.7-flash/",
      "metric": "pass^3",
      "version": "1.1",
      "date": "2026-05-29"
    },
    {
      "name": "Artificial Analysis Coding Index",
      "score": 37.1,
      "source": "https://openrouter.ai/stepfun/step-3.7-flash/benchmarks",
      "metric": "index",
      "date": "2026-06-15"
    },
    {
      "name": "SciCode",
      "score": 40,
      "source": "https://openrouter.ai/stepfun/step-3.7-flash/benchmarks",
      "metric": "percent correct",
      "date": "2026-06-15"
    },
    {
      "name": "Terminal-Bench Hard",
      "score": 35.6,
      "source": "https://openrouter.ai/stepfun/step-3.7-flash/benchmarks",
      "metric": "success rate",
      "date": "2026-06-15"
    }
  ],
  "release_date": "2026-05-29",
  "open_weights": true,
  "weights": [
    {
      "label": "Hugging Face",
      "url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash"
    }
  ]
}

Infrastructure

Provider availability

All providers

Provider	Endpoint	Context	Input /M tokens	Output /M tokens	Cached input /M tokens	Cache write /M tokens	Uptime 1d	Data handling
Novita	novita/fp8	262K	$0.20	$1.15	$0.04	—	100.00%	ZDR: Yes Training: Unknown Prompt logging: Unknown Retention: Unknown ZDR source Privacy policy Checked 2026-10-15
StepFun	stepfun/fp8	256K	$0.20	$1.15	$0.04	—	100.00%	ZDR: No Training: Unknown Prompt logging: Unknown Retention: Unknown ZDR source Checked 2026-10-15

Provider data:OpenRouter

Deployment graph

Can I run this?

Vertical Tabs only shows compatibility when it has structured evidence tying this exact model to hardware, runtime, quantization, memory use, and observed performance.

Browse hardware