FLUX.2 model family
Flux2 AI gives you the full FLUX.2 catalog from Black Forest Labs-choose the model that matches your balance of fidelity, speed, licensing, and hardware budget. See our features and use cases for applied examples.
What's new (2026)
- March 3, 2026 - FLUX.2 [pro] upgrade: Faster renders and sharper text/layout fidelity.
- Jan 29, 2026 - FLUX.2 [flex] speed bump: Around 3x lower latency with FP8 optimizations.
- Jan 15, 2026 - FLUX.2 [klein] launch: Sub-second 4B/9B models for interactive drafts.
- 4MP + 10-reference control: Keep characters, products, and layouts consistent across batches.
FLUX.2 [max]
Highest fidelity
Top-tier 4MP generations and edits with maximum coherence and detail. Choose [max] when print quality and visual polish matter most.
Key Features:
- 4MP outputs: Native high resolution for cropping and print.
- Layout adherence: Best at structured scenes and typography.
- Multi-reference: Keeps subjects consistent across sets.
- Precision editing: Strong for retouching and in-painting.
Best For:
- Print ads and OOH campaigns
- Packaging, product hero shots
- Editorial/brand visuals needing pixel detail
FLUX.2 [pro]
Production balance
Reliable, production-ready quality with the March 3, 2026 speed/clarity upgrade; strong text rendering and color stability for daily campaigns.
Key Features:
- Balanced quality/cost: Optimized for daily workloads.
- Clean typography: Better posters, UI mocks, and labels.
- Multi-reference: Consistent subjects and brand assets.
- Good latency: Faster than [max] while staying sharp.
Best For:
- Digital campaigns and social ads
- Website creative and landing pages
- Batch production of consistent visuals
FLUX.2 [dev]
Open weights
Open-weights variant with 4MP quality and multi-reference control-ideal for customization, research, or on-prem deployment.
Key Features:
- Open weights: Self-host or fine-tune within license terms.
- Up to 10 refs: Consistency for characters/products.
- 4MP editing: High-res in/out for retouch workflows.
- Great text/layout: Handles logos, UI, and posters.
Best For:
- Research and custom pipelines
- On-prem or private cloud deployments
- Teams needing deeper control over prompts and safety
FLUX.2 [flex]
Speed/quality tuning
Balanced model you can dial for latency or fidelity; now roughly 3- faster after the January 29, 2026 performance update.
Key Features:
- Tunable steps: Trade speed vs. detail per job.
- Good typography: Keeps UI and posters readable.
- Efficient runtime: Lighter VRAM footprint with FP8.
- Multi-reference: Keeps subjects aligned across shots.
Best For:
- High-volume production batches
- AB testing creative variants
- Realtime previews that still need quality
FLUX.2 [klein]
Interactive speed
Lightweight 4B/9B family designed for sub-second drafts and interactive prompt exploration without sacrificing coherence.
Key Features:
- Sub-second drafts: Great for live ideation sessions.
- Low VRAM: Friendly to smaller GPUs and edge.
- Reference support: Maintain characters across quick iterations.
- Upgrade path: Hand off final renders to [pro]/[max].
Best For:
- Interactive art direction sessions
- Rapid storyboard/thumbnailing
- Hardware-constrained deployments
Model Comparison
| Model | Speed | Quality | Best For |
|---|---|---|---|
| FLUX.2 [klein] | 5/5 | 3/5 | Interactive drafts |
| FLUX.2 [flex] | 4/5 | 4/5 | Volume + balance |
| FLUX.2 [pro] | 3/5 | 5/5 | Campaign creative |
| FLUX.2 [max] | 2/5 | 5/5 | Print & premium |
| FLUX.2 [dev] | 3/5 | 4/5 | Custom/self-host |
Choosing the right model
Match the model to your goal:
- Need realtime ideation? Start with [klein] then upscale in [pro] or [max].
- Need balanced throughput? Use [flex] for batches and A/B tests.
- Need top fidelity? Choose [max] for print and premium creative.
- Need control/self-hosting? Go with [dev] to customize pipelines.
Still unsure? Review our use cases or contact the team for tailored guidance.
Ready to start generating?
Pick a FLUX.2 model, plug in your references, and go. View our pricing plans to get started.