Haiku, Qwen, GPT-5 Mini and MiniMax refreshed
174 successful model/prompt builds published in autonomous sessions with browser tools. See each run for its output, transcript, runtime and cost.
This refresh includes Claude Haiku 5.5, Qwen 3.8 Flash, Qwen 3.6 35B A3B, GPT-5 Mini, and MiniMax M3. MiniMax’s Arpeggiator run remains on its previous output after a container failure; its new result is excluded from this release.
Top builds in this update
OUTPUT claude-haiku-5.5 Top-down alien shooter
Generated pi 0.87.1 · browser iteration · 100.0s · $0.0139 · 13k out
Full run + transcript →
OUTPUT qwen3.8-flash Top-down alien shooter
Generated pi 0.87.1 · browser iteration · 681.8s · $0.0593 · 39k out
Full run + transcript →
OUTPUT qwen3.6-35b-a3b Top-down alien shooter
Generated pi 0.87.1 · browser iteration · 72.6s · $0.0059 · 7.3k out
Full run + transcript →
OUTPUT gpt-5-mini Top-down alien shooter
Generated pi 0.87.1 · browser iteration · 75.8s · $0.0125 · 8.3k out
Full run + transcript →