FLUX.2 [max]

The guys from BFL have released their most powerful model! It is better than FLUX.2-flex and FLUX.2-pro, but inferior to Nano Banana Pro in text2image and several other models in image editing. * **Grounded generation** — searches the internet for relevant context in real-time: results of yesterday's matches, current weather, historical events with precise details. Just add the phrase "Search the internet" to your request — and the magic will begin. * Up to **10 reference** images. Products, characters, and styles remain consistent. On the leaderboard lmarena.ai, the model ranks 3rd and 7th: 🔹 #3 in Text-to-Image 🔹 #7 in Image Edit And on the leaderboard artificialanalysis.ai: 🔹 #2 in Text-to-Image and in Image Edit Available only via API, and loras cannot be trained either. __Share your tests in the comments.__