OpenAI's research agent for deep competitive and creative analysis lands in FLORA.

🤖 New Models

• o3 Deep Research – OpenAI's autonomous research agent for comprehensive multi-source analysis. Searches, interprets, and synthesizes hundreds of online sources—text, images, and PDFs—to produce analyst-grade reports with full citations. The model plans multi-step research trajectories and pivots based on what it discovers, backtracking when needed. In FLORA, use it to kickstart creative briefs with competitive landscape analysis, gather reference material across dozens of sources, or research trends before building your workflow. What used to take hours of tab-juggling now feeds directly into your canvas.

🧠 Why Deep Research in FLORA?

Creative work increasingly starts with research—understanding what exists, what competitors are doing, and what references to draw from. o3 Deep Research lets you front-load that discovery phase directly in your workflow. Ask it to analyze visual trends in a category, compile brand positioning across competitors, or gather technical specifications for a product shoot. The output flows naturally into your creative process, giving you a research foundation without leaving FLORA.

Three major image and video models land this release, each pushing the boundaries of what's possible in a single generation. Plus, optional two-factor authentication for teams that need tighter account security.

🤖 New Models

• GPT Image 1.5 – OpenAI's flagship image model with dramatically improved editing precision. Up to 4x faster than its predecessor with best-in-class instruction following. Excels at targeted edits that change only what you ask for—adjusting lighting, swapping outfits, or refining expressions—while preserving everything else. Significantly improved text rendering handles denser, smaller type for mockups and marketing materials.

• Wan 2.6 – Alibaba's cinematic video model with multi-shot storytelling and synchronized audio. Generate up to 15 seconds of 1080p video with lip-synced dialogue, sound effects, and music in a single pass. Reference-driven generation keeps characters, voices, and styles consistent across shots—ideal for ads, product videos, and narrative content without post-production.

• Flux 2 Max – Black Forest Labs' top-tier image model for maximum quality and control. The most capable FLUX variant yet, with the strongest prompt adherence and highest editing consistency in the lineup. Reference up to 10 images to maintain character identity, product details, and brand elements across batches. Outputs at up to 4MP with photorealistic lighting and physics—perfect for product imagery, marketing assets, and any work where consistency matters.

📦 New

• Two-factor authentication – optional 2FA for stronger account security. Enable in your account settings for an extra layer of protection on your workspace.

• Replace image in toolbar – swap assets directly from the block toolbar. Faster iteration when you need to update a source image without opening menus or rewiring connections.

Give your team access to AI generation without giving away the budget. Set per-user credit limits directly from Admin, with full visibility into who's using what — perfect for agencies managing client access or teams scaling up without surprises.

📦 New

• Per-user credit limits – Set spending caps for any user from Settings → Workspace → People & Credits. Choose one-time limits for project-based work or recurring limits that reset monthly. Your team gets creative freedom; you keep budget control.

• Usage visibility – See credit consumption at the user level, so you always know where resources are going before the bill arrives. No more end-of-month surprises.

Every block now remembers its history. Browse, compare, and reuse previous outputs without cluttering your canvas or losing that perfect variation from three iterations ago.

📦 New

• In-block generation history – Every block stores up to 100 past generations. Scroll through them in a carousel, navigate with keyboard shortcuts or trackpad gestures, and duplicate or export any result from the new menu.

• Default variation behavior – Choose how variations work for you. Head to Project Settings → Preferences → Variation Behavior to set whether new generations appear on canvas or stay in-node by default. Switch anytime based on how you like to work.