Highly impressed by Muse Image! It beats Nano Banana 2 across 7 tests, tho GPT Image 2 is still the champion
We tested GPT Image 2, Muse Image, and Nano Banana 2 across 7 dimensions: character consistency, poster design, prompt adherence, realistic rendering, storyboard, style
1/ releasing muse image today — the first image generation model from MSL. it's agentic: pairs with muse spark to reason through your prompt, search the web, and plan before it generates. people get what they meant on the first try. live now in the Meta AI app.
Lately I've been getting questions in my dm like, "Yo 3ench what's GMI_CLOUD?"
Today, I'll try to break it down the best way i can for everyone new to it.
we now have another open source model that is great with design and as affordable as DS V4!
tested Tencent Hy3, GLM 5.2, GLM 5.1, and DeepSeek V4 using three different webpage design prompts (all medium effort).
- Visual Output: Hy3 outperformed both GLM 5.1 and DeepSeek V4.
-
Café GMI is here! 1st stop: Seoul, Korea.
July 8, if you’re around @icmlconf, come grab a coffee and meet us at our café near COEX.
We’ll be there from 10 AM–6 PM with @Sky9Capital , AttentionX, and @ZZZZZPotentials
See you here: luma.com/gmicloud-2mqn?…
The shining depths of the ocean open the way to a world where legends come to life and incredible beauty blooms.
Created in @gmi_cloud
Model: Seedance 2.0
Prompt ⤵️
Claude Fable 5 totally crushed our July 4th animation contest, but it cost nearly 8x more than Opus 4.8, and GLM 5.2 is the cheapest!
We gave Fable 5. Sonnet 5, Opus 4.8, and GLM 5.2 the same tasks to test their motion graphic and animation outputs in celebration of Independence
