Fortytwo launches Engine API, reporting 24.7% fewer tokens on coding tasks

Fortytwo has launched Fortytwo Engine, an inference API where cooperating models steer a shared reasoning trajectory. Across 450 DeepSWE runs, Qwen 3.8 27B used 24.7% fewer tokens (12.3M vs 16.3M per run) and scored 47.8 versus 42.2 for the base model.

• Token usage fell from 16.3M to 12.3M per DeepSWE run
• Score rose from 42.2 to 47.8 (+5.6 points)
• Totals: 5.6B tokens for Engine vs 7.4B for base model
• API is OpenAI Chat Completions-compatible at base-model pricing