The Denoisies: 2025's Best and Worst in AI

Over on Denoised, we took a look back at the year with our first-ever Denoisies, handing out 13 AI awards covering everything from Best Image Model to Funniest AI Fail.

A few highlights:

Best Image Model:

• Addy's pick: Z-Image - The lightweight, locally runnable model wins for its customization via LoRAs.

• Joey's pick: Nano Banana Pro - Joey went with Google's model for pure efficiency in getting the best results as fast as possible.

Best Video Model:

• Joey's pick: Veo 3 - As the most definitive model of 2025, Veo 3 dominated in detail rendering and prompt adherence, setting the standard for the year.

• Addy's pick: Kling 2.6 - Addy crowned it the "people's champ" for ending the year neck-and-neck with Veo 3.1 despite coming from smaller company Kuaishou.

Biggest AI Comeback:

• Addy's pick: OpenAI Sora - Dominated the zeitgeist early in 2024, disappeared while competitors emerged, then came roaring back with Sora 2 and the meme-creation app that became the number one download that week. Despite eventually neutering the fun IP generation features, the comeback was undeniable.

• Joey's pick: Google - Went from irrelevant video models at the start of 2025 to dominating with Veo 3, Nano Banana Pro, and Gemini across nearly every category.

Best Open Source Release:

• Both agreed: Wan 2.2 Animate - Alibaba's model won for simplifying ComfyUI workflows and enabling real-world puppeteering techniques. Think Honey, I Shrunk the Kids made today—shoot live footage with puppets or sticks, use SAM 3D to extract the puppeteer tools, and let Wan animate and relight the environment naturally. The workflows went from complex multi-step processes to streamlined, practical VFX tools.

Biggest Drop-Off in AI:

• Addy's picks: Leonardo AI and Midjourney - Leonardo, once Addy's first AI subscription with iOS app firsts and real-time generation, has gone quiet. Midjourney, despite having a video model, seems stuck on diffusion-based architecture while competitors moved to multimodal, world-understanding systems.

• Joey's picks: Pika and Krea - Pika pioneered VFX models where you could add elements to existing footage, but hasn't capitalized on its real-time video potential. The lack of API access and architectural updates left both behind as the industry moved forward.

Most Overhyped AI Story:

• Both agreed: Tilly Norwood - The so-called "AI actor" that ruffled industry feathers with ridiculous trailer footage and overblown PR coverage from Variety, THR, and Deadline. "Anyone that works in AI knows how easy it is to create a synthetic human and keep consistency. To make a Tilly Norwood would take me like 10 minutes," Addy said. Joey added: "Hats off to whoever the hell their PR agent is 'cause they did a killer job spinning a non-story into this ridiculous story."

Most Under-Hyped:

• Addy's pick: Netflix research papers - Papers like Go-with-the-Flow signal what's coming down the pipeline from major studios. "Netflix is the biggest mover in the industry—if they're doing cutting-edge research on AI, we should be looking in that direction because that's what's coming, whether you like it or not."

• Joey's picks: ComfyUI's explosive growth - Went from nerd researcher tool to everyday production tool for serious creators, cementing itself as the Unreal Blueprints of AI. 

Biggest Acquisitions and Deals:

• Joey's pick: Figma acquiring Weavy - Bringing a big AI player into an established design tool felt more impactful than Adobe-Invoke. Figma's agility and newer architecture means better chances of actually integrating AI value versus Adobe's legacy rigidity.

• Addy's Pick: Disney and OpenAI Partnership - A major signal, though he questions if a blanket partnership limits Disney's diverse creative needs

Most Surprising Partnership:

• Addy's pick: Suno and Warner Music Group - Going from "most hated" to shaking hands through what's likely a revenue-share model giving Suno access to Warner's training data. A stunning turnaround for a company that was in lawsuits from day one.

• Joey's pick: Disney partnering with OpenAI on Sora - Especially notable since Disney sued Google for IP on the same day they announced the OpenAI deal. 

Biggest Missed Opportunities:

• Addy's pick: OpenAI not capitalizing on Sora app success - The app was the number one download with millions using it, but they neutered the fun IP generation features. "They should have went after licensing rights and just paid Disney for The Mandalorian generations. Just start paying IP for IP so the platform gets bigger."

• Joey's pick: Shutting up about your AI spot - Too many companies over-promoted their AI process, turning audiences off. Netflix did it right with The Eternaut, revealing AI use months after release when people already watched and formed opinions.

Biggest Surprises of the Year:

• Addy's pick: Nano Banana Pro's capabilities - The level of text adherence, prop conformity, preserving details while changing just one element, diversity in rendering people, frame composition, and camera language. "I was not expecting that this year."

• Joey's pick: Claude Code - AI that runs locally on your computer and actually manipulates files and rebuilds things, enabling non-coder people to spin up functional applets. The direction of making software development accessible to everyone.

Biggest Disappointment:

• Both agreed: GPT-5 - Overpromised and underdelivered, with models that are "confidently incorrect" across the board. 

Funniest AI Fails:

• Addy's pick: McDonald's commercial - The terrible AI-generated Christmas ad that got pulled after massive backlash.

• Joey's pick: ChatGPT-generated prayer - Read at church, included "feel free to modify this prayer or tell Lord" in the actual prayer.