Flux.1 Dev
Flux.1 Dev is Black Forest Labs' 2024 open-weight stills sibling of Flux.1 Pro, not the live flagship and not a fully MIT gift. Rechecked 2026-08-17: bfl.ai now leads with FLUX 3. Current stills and open-weight rows sit on FLUX.2 ([dev] / [klein]) at bfl.ai/models/flux-2 and bfl.ai/pricing. github.com/black-forest-labs/flux is the official FLUX.1 inference repo: 25,892 stars, GitHub SPDX Apache-2.0. Drop leftover fully open source / Pro-level quality as if it were MIT forever and still the current default.
Compare FLUX 3 if you wanted the live multimodal model, Flux.1 Pro if you wanted the 2024 paid sibling, and GPT Image 2 if you wanted OpenAI stills.
Key Features
- Previous open-weight stills: Flux.1 Dev shipped in August 2024 next to Flux.1 Pro. Recheck whether your workflow should move to FLUX.2 [dev] or [klein].
- License is not "do anything": BFL's live pricing sells open-weights license packs (Builder / Platform / Professional / Enterprise / Synthetic Data). Do not call Flux.1 Dev a fully commercial MIT dump.
- Inference repo vs model card: The flux GitHub repo is Apache-2.0 inference code. Model weights still have BFL's own non-commercial / licensed terms. Recheck Hugging Face and the self-hosted terms.
- Not the homepage model: Image generation on the 2026 homepage is FLUX 3 (soon) plus FLUX.2.
Use Cases
- People with a 2024 Flux.1 Dev ComfyUI workflow who need the catalog corrected.
- People comparing leftover Flux.1 Dev vs FLUX.2 [dev].
- People who thought the weights were fully MIT.
Limitation: We did not re-download the 2024 weight card. Recheck Hugging Face and BFL self-hosted terms before you ship commercially.
Pricing
bfl.ai/pricing and GitHub on 2026-08-17.
| Piece | Price | Notes from those pages |
|---|---|---|
| Flux.1 Dev weights | Previous open-weight SKU | Recheck Hugging Face / self-hosted terms. Not a MIT commercial dump. |
| FLUX.2 open weights | License rows | [klein] / [dev] packs on the live pricing page. |
| flux inference repo | $0 code | Apache-2.0. 25,892 stars. |
If a leftover blog still says Flux.1 Dev is fully open source and still the BFL default, treat this table as the source.
Getting Started
- Open bfl.ai/models/flux-2 before you download 2024 Dev weights.
- Recheck FLUX 3 if you wanted video and audio.
- Recheck Flux.1 Pro if you still have an old paid key.
- Do not call it fully MIT.
Frequently Asked Questions
Is Flux.1 Dev still the open-weight flagship?
No. Live open-weight rows we opened are FLUX.2 [dev] / [klein].
Is it fully open source?
The inference repo is Apache-2.0. Weights are under BFL's own terms. Recheck before commercial use.
Same as Flux.1 Pro?
Sibling 2024 SKUs. Pro was the paid API stills model. Both are previous catalog.
Alternatives
- FLUX 3: Live multimodal BFL model.
- Flux.1 Pro: 2024 paid stills sibling.
- GPT Image 2: OpenAI stills.
Tips
- Quote FLUX.2 [dev]/[klein] as the live weight rows.
- Split inference-repo Apache-2.0 from weight terms.
- Recheck self-hosted terms before you sell a commercial fine-tune.
Conclusion
Flux.1 Dev is a 2024 BFL open-weight stills SKU, not a fully MIT flagship. Start at bfl.ai/models/flux-2, then decide whether FLUX 3 already covers the model you need.
Comments
No comments yet. Be the first to comment!
Related Tools
Flux.1 Pro
bfl.ai/models/flux-2
Previous BFL image SKU. Rechecked: live catalog is FLUX 3 and FLUX.2. Not a 2024 Midjourney/DALL-E 3 crown.
Stable Diffusion 3.5
stability.ai/stable-image
Stability 2024 stills family. Rechecked: community license, not fully MIT. Homepage also leads with audio and Brand Studio.
OpenAI: dall-e-2
developers.openai.com/api/docs/models
Previous OpenAI DALL-E 2 snapshot. Rechecked: live image path is GPT Image 2. DALL-E 3 is already removed from the API.
Related Insights

Grok Bot and Hermes Bot: one person finally gets a think tank and a secretariat
Grok Bot now ships with Cursor Pro+. Hermes Bot runs on a VPS. They are not smarter chat boxes. The think tank advises, the secretariat executes, and you still make the call.
Hook OpenCode Go into Codex on Windows. Do Not Open a Second Toolkit.
A ChatGPT-signed Codex desktop app still shows mostly GPT in the picker. On Windows, enable only OpenCode Go and the Grok, GLM, Kimi, DeepSeek, and MiniMax models you already pay for appear in the same selector. Keys stay local. Native GPT stays put.
What Locks Codex Is the Picker, Not the Models
You already pay for OpenCode Go, Grok, and Z.ai, but the Codex picker still shows mostly GPT. The community project codex-router does not teach another install ritual. It puts subscriptions you already bought back into the selector, keeps keys on the machine, and leaves native GPT alone.