Anthropic's Fable 5 generates video games and complex maps from single prompts, researcher finds

Anthropic's Fable 5 can make weirdly fun video games with the click of a button

Anthropic has released Claude Fable 5, the first publicly available version of its closely watched Mythos model. What can Fable actually do? All kinds of things, it turns out. Ethan Mollick, a notable AI researcher and an associate professor at Wharton, has been playing around with the model and seems to be having a lot of fun. In his testing, Fable consistently "outperformed basically every other public model I have used by a considerable margin," Mollick wrote Tuesday on his Substack. He added that it was "capable across many problems and produced some startling results -- it would work up to a dozen hours executing on multi-page specifications." Perhaps most strikingly, Mollick used Fable to create a variety of video games -- all of which were generated via "one initial prompt" in Claude Code, the researcher says. Among these, Snake is exactly what it sounds like. You're a Pac-Man-like snake and you roam around eating apples. The snake never stops moving, and if you run off the screen, you die. It's very 1980s arcade but, like many of those old games, it's weirdly addicting. I played it longer than I'd like to admit before remembering I am a gainfully employed writer and not, in fact, a serpent who likes fruit. Then there was Strata, where you're roaming around in a seemingly endless network of subterranean tunnels and the goal is just to light as many lanterns as possible. The graphics look like a degraded version of Myst -- they aren't great -- but the fact that the game exists at all, generated from a single prompt, is impressive. Mollick even managed to create Duino, a game based on the Duino Elegies, the celebrated cycle of poems by German poet Rainer Maria Rilke. I like the animation here best -- the player is a lone figure in a nocturnal landscape -- although there isn't much to the gameplay other than walking around while Rilke passages materialize on the screen. Aside from the variety of instant games Mollick produced, he also used Fable to create an isochronic map -- a visualization showing how long it takes to travel between any two locations. The accuracy and detail is arresting. The implications are pretty clear. Software projects that once required entire teams -- games, mapping tools, highly complex specifications -- are now being spun up from a single prompt. It's reason for vibe coders of the world to rejoice. As for founders and operators watching AI capability curves, it's a useful data point about how quickly the floor is rising.

Did Anthropic Just Shift the AI Narrative With Fable 5? AI Researcher Ethan Mollick Thinks So

The research suggests that AI can solve problems involving research, math, visual development, taste, judgment, complex coding and much more Having read through the detailed blog post on Claude Fable 5 shared by Anthropic, questions raced one another within our team. However, that's when someone pointed us in the direction of Ethan Mollick, a well-known AI researcher and scholar at the Penn university. He says he's been playing around and having lots of fun with the new model at making funny video games. Our first response was Phew! Now, here's something that that we can actually use AI for. No, that's not entirely true given that we've been using it to create whacky pictures that you may have observed from time to time with our posts on this website. However, before going to the good things, Mollick shares some facts about Fable 5, the Mythos Class AI model to which he got early access. While noting that the discussion has centred on software security (our team's internal questions took that route), he says the guardrails around Fable 5 prevent it from being used for cybersecurity at all. "My conclusion is that it represents a very real leap over every model I have used before, and, maybe more important, suggests our relationship with AI is changing in drastic ways," he writes in a post on Substack. He further noted that Fable "outperformed basically every other public model I have used by a considerable margin" and appeared to be "capable across many problems and produced some startling results -- it would work up to a dozen hours executing on multi-page specifications." It is in this context that Mollick used Fable 5 to create a few video games - all generated with one initial prompt in Claude Code. The games include Snake, Strata and Duino with the first one doing what the old Pac-man would do by roaming around and eating apples. It never stops moving and in case one runs off the screen, one dies. "It is very 1980s arcade, but like many of those, it is weirdly addictive," he said. The researcher explains that Strata is a game where the player is roaming around in a seemingly endless network of tunnels with the goal being to light as many lanterns as possible. The graphics aren't great but are functional. Then there is Duino which is based on the celebrated cycle of poems by Rainer Maria Rilke where the player is a figure in a nocturnal landscape who walks around as the poems appear on the screen. While these games were fun and appeared juvenile (well, that's okay isn't it?), what got us excited about Fable 5 was its ability to create an isochronic map which is a visualisation of how long it can take to travel between two locations. Yes, Google Maps does it all the time, but this one is interesting when you see the accuracy and details it adds. According to Mollick, "no previous model did an even halfway useful job with trying to create a map like this because it involves researching thousands of potential trip distances and a lot of small judgement calls and decisions." He also provided a transcript of the building session that Fable 5 went on its own and even launched multiple other Ais to help it with tasks such as researching on travel times on flights, rail schedules etc. The result was a fully functioning map of impressive sophistication that looked a lot like the 1881 original, but that doesn't mean it was perfect. I noticed that a lot of remote locations (like Greenland) just contained estimates of travel time, not exact numbers, he said while underscoring that this was "probably not a useful project." However, what Mollick has pointed us to is rather indicative of how AI can solve problems involving research, math, visual development, taste, judgment, complex coding and much more. "And the unnerving part was how little I did. I gave a really ambitious instruction, the AI followed it. I gave a couple of minor pieces of feedback, and the AI figured it out. My role was extremely limited," he concludes.