Square Enix has released a curious improvement for its MMORPG Dragon Quest X — Google’s Gemini artificial intelligence has been officially introduced into the game. The technology formed the basis of a new satellite named Oshaberi Slimi, which can communicate with users in real time by voice or via chat. The release of this feature is scheduled for the near future, and the AI assistant itself is already able to analyze what is happening on the screen and give practical advice on passing.
Table of Contents
Fuck Slimy with Gemini: how AI improves the gaming experience in Dragon Quest X
The developers from Square Enix decided to make the gameplay of Dragon Quest X friendlier, especially for beginners who find it difficult to get used to the vast world. Oshaberi Slimy is not just a decorative pet. Thanks to Gemini integration, he understands the context of the current situation and can react to the player’s actions literally on the fly. In fact, this is an interactive guide that is always at hand.
Interestingly, AI is able to maintain a full-fledged dialogue. You can ask him for directions or specify tactics for the boss — the system will process the request and give you an answer without straying from the style of the fantasy universe. Such experiments with neural networks in games are gradually becoming mainstream, but Square Enix is one of the first to implement Google’s solution in such a large—scale project.
The choice of a specific neural network for such an experiment was dictated not only by marketing, but also by pure technical calculation — Square Enix needed a solution capable of “seeing” the game through the eyes of a gamer.
Google Gemini vs GPT-4: Why Square Enix Chose Multimodal AI
Google Gemini is a whole family of multimodal models that can work with text, sound, and images out of the box. This is their main advantage over the same GPT-4: there, multimodality was gradually tightened, and Gemini was originally designed as a single native architecture. This structure allows Chatty Slimey to analyze what is happening in Dragon Quest X in real time — the AI instantly recognizes bosses like the Nirvana King or rare loot like Dragontail, giving contextual advice. By the way, the capabilities of the Gemini 1.5 Pro version are impressive: the model holds up to 1 million tokens in memory — this volume is enough to memorize the entire colossal lore of the Astoltia world.
The partnership with Google Cloud was concluded for the sake of phenomenal performance: Chatty Slimey uses Gemini Pro for Japanese speech synthesis, maintaining a delay of around 200 ms even under peak loads on DQX servers. Apparently, the Japanese were also bribed by the model’s leadership in MMLU benchmarks (accuracy above 90%), where it outperforms competitors in complex logical tasks — an ideal tool for developing tactics for raids from the latest version 7.0. Plus, integration through the Vertex AI platform minimizes the load on the user’s hardware. All the “intelligence” revolves in the Google cloud, and not on consoles or PCs of players — this is critically important for MMORPGs with online more than 300k+ active users per month, where any extra load can turn into lags.
Dragon Quest X: 12 years of MMORPG Development and the first Gemini implementation
For those who missed it: Dragon Quest X is a long—running MMORPG from Square Enix, launched back in 2012. The project successfully combines the classic mechanics of the legendary series with online elements: world exploration, quests, and social interaction. The introduction of AI is perhaps the most serious technological leap for the game in recent years.
By the way, while Square Enix is improving the availability of its “top ten”, other developers are teasing the dangers in their new products. So, Bungie studio recently revealed details about Marathon — the creators of the action shooter officially named the most dangerous place on the map. Against the background of such news, the world of Dragon Quest X looks much more comfortable, especially if smart Oshaberi Slimy is hovering behind you.
