Claude Opus 5.5 burns $7 to make a cow that runs, “What is this nightmare?”
This is what AI companies have been warning us about.
Cow jumping GIF by Cybernews/Claude Opus 5.5 via AiMind on X
Follow us
- An X user paid $6.80 to have Claude Opus 5.5 generate a cow running.
- The model produced a cow running on udders, drawing mockery and attention online.
- The post reached more than 921,000 views in under 24 hours.
- Anthropic markets Opus 5.5 as its strongest model, but it is not built for video generation.
Key Takeaways by nexos.ai, reviewed by Cybernews staff.
AI functions at extremes. While many of us fear being tricked by convincing deepfakes, others are dumbfounded by how wrong AI can be.
We live in a dystopia where romance scammers trick the elderly into sending money to “Brad Pitt,” and state-sponsored activists generate copies of political figures to stoke the fires of social unrest.
This is largely thanks to AI and big tech companies like OpenAI and Anthropic, who have made this nightmare a reality.
But where there are sophisticated AI-generated images and videos, there are outright ridiculous creations that defy the laws of physics.
In Claude’s world, cows run like this
One X user was shocked after experimenting with Anthropic’s cheapest and strongest release and asked the model to generate a “cow that runs.”
This version generated a cow with udders for its legs.
“It runs on its f*cking udder. Four teats as legs. Then it jumps a fence in slow motion and lands like a pancake,” the user describes.
The animation also cost the user $6.80.
But it seemingly wasn’t a waste of money, as this “running” cow captivated the hearts of nearly 1,000 X users with its physics-defying physique.
The post has also amassed over 921,000 views in under 24 hours.
“What is this nightmare?” one user asked, while others agreed that “this is so much better than what you asked for.”
Another user asked sarcastically, “Is this an actual video of a cow running, or is this the AI one?” poking fun at the deepfake dilemma social media users find themselves in these days.
While people argued that the output may be a result of Claude’s “personality,” it's clear that Anthropic is slowing down in terms of AI advancement.
New meaning to “pacing the frontier”
Anthropic announced Claude Opus 5.5 straight off the back of Dario Amodei’s essay We Must Pace the Frontier.
In this essay, the Anthropic CEO said,” We must slow the pace at which we improve the capabilities of AI models” because of AI’s recursive self-improvement and the OpenAI-Hugging Face incident.
The conversation on this topic is live. Join in the discussion.
While Amodei talks of pacing AI advancement to avoid further harm, Anthropic has released Opus 5.5, marketed as “the strongest-performing model” the company has tested thus far.
The company likens it to its infamous Mythos 5.1 model in biology and cybersecurity, with versions of Mythos dubbed too risky for public release.
While it boasts impressive cybersecurity, deep knowledge, and coding capabilities, Anthropic’s Opus 5.5 seems to be pacing its models in AI image and video generation.
Although Claude Opus 5.5 isn’t designed for animation, video, or image generation, the company said that during testing, someone built a “game from a single prompt.”
“Opus 5.5 scored higher than any other model on the strength of its graphics and polish.”
