GPT-6.1 Terra builds 3-level game in 6.5 min; DeepSeek V4.1 Flash slower but 36% cheaper
latent_lurker8587 · reddit · 2026-11-01
A Reddit user benchmarked GPT-6.1 Terra against DeepSeek V4.1 Flash on the same task via Atomic Agent's Fusion mode: a code-drawn, 3-level Halloween witch game with a self-playing demo mode.
- GPT-6.1 Terra: 14 agent steps, 6:29, $0.44, 91% cache hits
- DeepSeek V4.1 Flash: 73 steps, 23:43, $0.28, 96% cache hits (kept high by Atomic Agent's stable prompt prefix)
- Terra was 4x faster; DeepSeek was 36% cheaper
A local Qwen 3.8 27B on a MacBook ran out of context and produced a blank screen. The full game prompt is included, with the closing question: 24 minutes for $0.28, or 6 minutes for $0.44?
More from coding & agent
- ASC CLI adds territory-scoped PPP pricing, deep IAP validation, serial matching — rudrank · 2026-11-01
- One-Command App Store Sale Scheduling: asc CLI Example With Auto-Revert Pricing — rudrank · 2026-11-01
- asc versions release Now Reconciles After Server Errors, So Shipped Releases No Longer Turn CI Red — rudrank · 2026-11-01
- asc CLI now reconciles release status after server errors to avoid false CI failures — rudrank · 2026-11-01
- asc CLI adds terminal-based streamlined purchasing toggle and StoreKit ownership filtering — rudrank · 2026-11-01
- asc CLI Gets Resilience Fixes: Serial-Number Tolerance and Secret-Mangled Key Recovery — rudrank · 2026-11-01