Jev plays Tetris
Every piece is dropped by one real Jev call — the board and the legal placements go to the API, Jev picks one, we play it. Watch the decision counter climb, each in milliseconds for a fraction of a cent. Press play.
Same decision API you can put in your own app — a typed choice/score/noul call in ~100–300ms. Get a key and build with it.
How Jev plays Tetris
Every piece is placed by a real Jev API decision. When a new tetromino appears, we work out its legal landing spots, shortlist the strongest few, and send the board plus those options to the Jev endpoint as a typed choice. Jev returns the placement it judges best — with a calibrated probability — in roughly 100–300ms, and the piece drops there. One genuine model call per piece, live.
The counters on the page are the same numbers a developer sees building on Jev: the decision count, per-move latency and running cost (at Jev's own $0.042 per million input tokens). Tetris just makes them watchable — you are seeing the API decide in real time.
What is Jev?
Jev is TypeSafe AI's first System One model. Instead of writing prose like a chatbot, it reads a situation and returns a typed, calibrated decision — a choice, a score, or a yes/no probability — in milliseconds, with no hallucinations. Placing a Tetris piece is exactly that shape of decision, which is why the same model that routes tickets or gates an agent's tool calls can play Tetris piece by piece.
It is a general decision engine, not a dedicated Tetris bot, so it will not always find the optimal drop — the point is watching a real API make fast, cheap decisions live.
FAQ
Is the AI really playing Tetris, or is it scripted?
It is really deciding. For each piece the board and the candidate placements are sent to the Jev API, and it returns a typed choice of where to drop. There is no hard-coded Tetris strategy — it is a general decision model picking a placement each turn.
Is Jev Tetris free to watch?
Yes — it runs in your browser with no API key and no signup. Each placement is a free, rate-limited call to the live Jev decision API.
How fast and how cheap is each move?
Each placement is one API call that typically returns in about 100–300ms and costs a fraction of a cent in input tokens. The page shows the live latency and running cost.
More: Watch Jev play 2048 · Play Gomoku vs Jev · All decision games · Jev API docs