OpenAI: GPT-6.1 Terra

OpenAI's GPT-6.1 Terra establishes itself as a versatile mid-tier model, delivering strong capability across diverse domains while offering competitive pricing and generally performing in the top tier for speed. Operationally, the model demonstrates exceptional reliability, achieving a flawless 100% success rate across benchmark testing with zero technical interruptions. Across evaluated categories, Terra's standout asset is instruction following, reaching an impressive 90.0% accuracy (94th percentile) on complex, multi-step conditional directives. It also exhibits high competency in advanced mathematics, scoring 95.0% accuracy (84th percentile), alongside dependable results in coding with 93.0% accuracy (71st percentile). For both math and coding evaluations, execution latency was notably brisk, placing in the upper percentiles for response times. Terra's core strength lies in combining elite directive adherence with consistent computational precision at an accessible price point. Its primary trade-off appears in multi-step deductive reasoning; while accuracy remained solid at 94.0% (69th percentile), response duration was comparatively sluggish, falling to the 25th percentile in speed. Nevertheless, GPT-6.1 Terra remains an efficient, highly dependable engine well-suited for everyday agentic workflows, programming assistance, and detail-intensive document processing.