← All workflows

image / Pipeline / Beginner

Flux.1 Dev — Text to Image

Generate images from text with Flux.1 Dev using separate diffusion, text encoder and VAE loaders.

Download workflow JSON ↓Read setup instructions
Generated output from the recorded Flux.1 Dev — Text to Image test
Actual output from the recorded RTX 5080 run. See test settings below.

What it produces

Generate images from text with Flux.1 Dev using separate diffusion, text encoder and VAE loaders.

HARDWARE & SETTINGS

TESTED on NVIDIA GeForce RTX 5080 - UNCLASSIFIED

This is a catalog estimate, not a tested minimum. GPU architecture, precision, resolution and offloading affect whether a run fits.

Output settings
512 × 512
Workflow version
v1.0
System RAM
62 GB (test system)
Test hardware
RTX 5080 · 16 GB
Generation time
10.05 seconds in this run
Peak memory
Not measured
Read the compatibility policy →

HARDWARE INTELLIGENCE LAYER

Will this run on my GPU?

Evaluate evidence-backed execution feasibility for your specific hardware.

TESTED ON RTX 5080 - UNCLASSIFIEDConfidence: HIGH

Directly Tested on NVIDIA GeForce RTX 5080

Current workflow hash matches the successful unclassified execution. Success applies to its recorded settings.

CLOSEST EVIDENCE

NVIDIA GeForce RTX 5080 (16GB VRAM)

Exact GPU model and current workflow hash; applies only to the recorded settings and software.

Measured: 10.05s runtime

MEASURED RUNTIME SPECS

Resolution:
512 × 512
Steps / Sampler:
20 (euler)
Runtime:
10.05s
Test Date:
2026-09-08

Hardware & Runtime Considerations:

  • Historical success: STANDARD versus OPTIMIZED was not recorded.

Execution history (1)

Success applies to the recorded settings. Failure is an outcome; STANDARD and OPTIMIZED describe execution settings. Historical runs without that distinction remain unclassified.

Hardware / dateProfile / outcomeWorkflow evidenceDetails
NVIDIA GeForce RTX 5080
2026-09-08T04:02:56.136Z		UNCLASSIFIED
SUCCESS · 10.05s		CURRENT
70d5e91baabd		Exact record · Output
Settings and notes
{
  "steps": 20,
  "cfg": 1,
  "sampler": "euler",
  "scheduler": "simple",
  "resolution": "512 × 512",
  "batchSize": 1
}

Historical run: STANDARD versus OPTIMIZED was not recorded. Original evidence preserved.

No comparable cross-GPU executions. Timing comparisons require matching workflow, inputs, model hashes, settings, software, precision and timing protocol.

Models & custom nodes

These model filenames are extracted from the downloadable graph, including subgraphs. Check each source’s license and access requirements.

ae.safetensors ↗ComfyUI/models/vae/
clip_l.safetensors ↗ComfyUI/models/text_encoders/
t5xxl_fp16.safetensors ↗ComfyUI/models/text_encoders/
flux1-dev.safetensors ↗ComfyUI/models/diffusion_models/

Custom nodes

The graph uses ComfyUI core nodes. Use a recent ComfyUI version that includes these node types.

Standard ComfyUI graph.

Custom node setup guide →

Installation & first run

  1. Install ComfyUI using the official requirements and installation documentation for your operating system and GPU.
  2. Download this workflow JSON. Use the listed model sources above and place the required files in the indicated model folders. Model folder guide →
  3. Load the JSON in ComfyUI. Resolve missing nodes and select installed model files in the loader nodes. Supply any required input image, video or audio.
  4. Review the graph’s settings before running. See the settings extracted below; a recorded test only establishes the specific configuration used in that run.
  5. Run the workflow and check the ComfyUI console if it fails. Keep the exact error text, software versions and your hardware details for troubleshooting.

Example prompt from the catalog

A beautiful fantasy girl with long curly silver hair and big blue eyes, wearing a white transparent fairy dress with lac

This prompt is a starting point, not evidence that the preview was produced by this workflow.

Explore the workflow

When a run fails

Missing model or a model name is not available

Compare the graph’s loader file names with your installed files and the dependency list. Check model folders →

Missing or unrecognized node type

Record the exact node type and check its source and version before installing extensions. Custom node guide →

GPU memory or execution error

The VRAM estimate is not a guarantee. Record the error, input dimensions, model precision and batch size. GPU error guide → · Workflow error guide →

Graph settings & required inputs

No Load Image input file is required by this graph.

View the graph’s saved controls
KSampler · node 31
[192044955347061,"randomize",20,1,"euler","simple",1]

Testing & source information

This exact download completed a local run on an RTX 5080 with 16 GB VRAM. This does not establish a minimum VRAM requirement.

Test date
2026-09-08
ComfyUI
0.33.0
PyTorch
2.11.0.dev20260211+cu128
Resolution / batch
512 × 512 / 1
Steps / CFG
20 / 1
Generation time
10.05 seconds, including model loading
Peak memory
Not measured
Download the test record →

Catalog version: v1.0. Original publication date is not recorded.

Graph source: 01.json ↗.

What a complete test record includes →

Related workflows