FLUX 2 Flex is Black Forest Labs' controllable text-to-image model. Unlike FLUX 2 [pro] (zero-config), Flex gives you adjustable guidance scale and inference steps for precise quality/speed control, plus enhanced text rendering.

How much does FLUX 2 Flex cost on Runflow?

$0.05 per megapixel on both input and output, rounded up to the nearest megapixel. A 1024×1024 image costs $0.05. A 1920×1080 image costs $0.10. Even a 512×512 rounds up to $0.05. Cold starts are not billed.

What makes FLUX 2 Flex different from FLUX 2 [pro]?

FLUX 2 [pro] is zero-config — it optimises automatically at $0.03/MP. Flex gives you manual control over guidance_scale and num_inference_steps at $0.05/MP, plus enhanced typography. Choose Flex when you need to fine-tune outputs or render text accurately.

What are the recommended inference steps?

10–20 steps for fast prototyping and previews, 25–28 for balanced production quality (default: 28), and 40–50 for maximum fidelity final outputs.

Can I use FLUX 2 Flex commercially?

Yes. FLUX 2 Flex is available under a commercial license. You can use it in production applications, SaaS products, and client work.

Does FLUX 2 Flex support text rendering in images?

Yes. FLUX 2 Flex has enhanced typography and text rendering capabilities — it produces more legible and accurate text in generated images compared to most other models.

Do I need to manage GPUs or infrastructure?

No. Runflow handles all GPU provisioning, scaling, and failover. You make an API call, you get back an image.

Black Forest Labs

FLUX 2 Flex API

FLUX 2 Flex by Black Forest Labs — adjustable inference steps and guidance scale for fine control. Enhanced text rendering. $0.05/MP.

flux-2-flex— paramsRectified Flow Transformer v2

Updated March 2026·Early access

Run the API How Sentinel controls quality

Price

$0.05/MP

Per megapixel, input + output

Cold Start

5–10s

Not billed, handled by Runflow

Quality Score

Coming soon

Sentinel benchmarks in progress

FLUX 2 model family by Black Forest Labs

Multiple FLUX 2 variants for different quality, speed, and control trade-offs — all on the same Runflow API.

FLUX 2 [pro]

Pro

Production-grade — zero-config, automatic prompt upsampling, $0.03/MP

FLUX 2 Flex

This model

Adjustable inference steps and guidance scale for fine control, $0.05/MP

FLUX 2 Max

Premium

Maximum quality variant — highest fidelity in the FLUX 2 family

FLUX 2 Turbo

Speed

Speed-optimised — lowest latency for real-time and preview workflows

Benchmarked for your use case

Every model on Runflow is scored per vertical — not generic leaderboard numbers. Scores come from domain-expert evaluation on real production prompts.

Use Case	Assessment	Notes
Portrait	Strong	Fine-tunable inference for optimal skin detail and lighting
Fashion	Strong	Good style versatility across artistic and photographic domains
Product	Strong	Accurate rendering with tuneable quality/speed trade-off
Creative	Strong	Enhanced typography and text rendering accuracy

Per-megapixel pricing

Quality	Output Size	Megapixels
512 × 512	0.26 MP → rounds to 1	$0.05
1024 × 1024	1 MP	$0.05
1920 × 1080	~2 MP	$0.10

$0.05 per megapixel on both input and output, rounded up to nearest megapixel. Cold starts (5–10s) not billed.

API Parameters

Full parameter reference for the flux-2-flex model endpoint.

Parameter	Type	Required	Description
prompt	string	required	Text description of the image to generate.
image_size	object \| string	optional	Output size. Presets: square_hd, square, portrait_4_3, portrait_16_9, landscape_4_3, landscape_16_9. Or custom {width, height}. Default: landscape_4_3.
guidance_scale	float	optional	How closely the model follows the prompt. Range: 1.5–10. Default: 3.5.
num_inference_steps	integer	optional	Number of denoising steps. Range: 2–50. Default: 28. 10–20 for drafts, 40–50 for final quality.
output_format	string	optional	Output format. Options: "jpeg", "png". Default: "jpeg".
safety_tolerance	string	optional	Content moderation strictness. 1 = most strict, 5 = most permissive. Default: 2.
enable_safety_checker	boolean	optional	Enable NSFW content filtering. Default: true.
seed	integer	optional	Random seed for reproducible outputs.
sync_mode	boolean	optional	Return media as data URI without storing in request history. Default: false.

Technical Specifications

ArchitectureRectified Flow Transformer v2

Guidance Scale3.5 (default) · 1.5–10 range

Inference Steps28 (default) · 2–50 range · 10–20 drafts, 40–50 final quality

Aspect Ratiossquare_hd · square · portrait_4_3 · portrait_16_9 · landscape_4_3 · landscape_16_9

Output FormatsJPEG · PNG

Text RenderingEnhanced typography and text accuracy

LicenseCommercial

Why run FLUX 2 Flex on the Runflow API

Flux AI models are available on many platforms. Here's what makes the Runflow image generation API different.

Full inference control

Adjust guidance_scale (1.5–10) and num_inference_steps (2–50) to fine-tune the quality/speed trade-off per request.

Enhanced text rendering

FLUX 2 Flex has improved typography and text rendering accuracy — ideal for generating images with legible signage, labels, or overlays.

Step-based workflow

Use 10–20 steps for fast prototyping and previews, 25–28 for balanced production runs, and 40–50 for maximum fidelity final outputs.

Advanced prompt features

Supports JSON-structured prompts for granular scene control, HEX colour codes, and @ symbol image referencing for multi-image workflows.

Sentinel quality scoring

Every output is scored by Runflow's Sentinel system before delivery, catching low-quality or off-prompt results automatically.

Built for these workflows

Real production use cases teams run on FLUX 2 Flex today.

Product marketing with text overlays

Finalise ad creatives with promotional copy, price tags, and CTAs baked directly into the image — preserving fine product details underneath.

Film titles, credits & static assets

Generate text-heavy visuals like title cards, credit sequences, and poster typography where every letterform needs to be sharp and legible.

Creative design platforms — text tier

Offer a specialised text-placement model as part of a pro or premium subscription tier, handling higher quotas for users who need reliable typography.

Daily-rotating food & deal promotions

Swap out offer copy, prices, and language on restaurant and delivery ad templates daily — instant localisation without a designer in the loop.

Explore other models

Same API, different strengths. Switch models with one parameter change.

Black Forest Labs

Frequently asked questions

Resources

API Documentation

Full endpoint reference, authentication, and response formats.

Sentinel Quality Evaluation

Automated quality gates for every image in your pipeline.

Start generating images with FLUX 2 Flex

Get API access in minutes. No GPU setup, no infrastructure to manage. Pair with Sentinel to control the quality of every AI-generated image you ship.

Run the API Talk to Founders