—ms
0 decisions · 0 returns
Jev PongAbly Labs
Here's what that does to a game of Pong.
Recorded run · Vercel (iad1) via Vercel AI Gateway
replay0.0 s
—ms
0 decisions · 0 returns
—ms
0 decisions · 0 returns
Anthropic
—ms
0 decisions · 0 returns
—ms
0 decisions · 0 returns
modeldecisions/savg msp95 msin first 12 s
JevTypeSafe AI
4.4022740047
Gemini 3.8 FlashGoogle
0.32316774383
Claude Haiku 4.5Anthropic
0.40248383582
GPT-5.6 SolOpenAI
0.283529103502
Four lanes, one game: same serve, same rules, same question, asked of a different model in each. One answer moves the ball one step, so every ball travels at its own model's answer time and nothing is skipped or sped up. Jev is a typed-decision model rather than a chat model, and Pong is the Atari game chat models handle worst, so a loop that won't wait is the fair place to see what that's worth.
The four lanes are deliberately the four everyone knows. Since recording them the same question has been put to five more models, including GPT-6 Astra and Claude Fable 5.1. Same result. See the numbers.