GPT-6.1 Terra earns its spot: medium and high tiers land on the cost-intelligence Pareto frontier
Balance- · reddit · 2026-10-28
A Reddit post shares Artificial Analysis' Intelligence Index vs. cost-per-task scatter: GPT-6.1 Terra (medium/high) sits on the Pareto frontier inside the "most attractive quadrant," with GPT-6.1 Sol tiers close behind. GPT-6 Astra and Claude Opus 5.5 (max with fallback) score highest but cost $10-$20+ per task, while Claude Haiku 5.5, DeepSeek V4.1 Flash and MiniMax-M3.1-Flash occupy the low-cost band. Takeaway: GPT-6.1 Terra has found its niche — near-frontier intelligence at mid-tier pricing.
More from Models
- GPT 6.1 Terra excels at CAD, now default model for free users — npew · 2026-10-28
- Small open model learns to say "ask a doctor", cutting errors from 46 to 13 — MaziyarPanahi · 2026-10-28
- Richard Socher's Recursive open-sources Flint: 40B MoE, 2.5B active, Apache 2, runs coding locally on Mac — altryne · 2026-10-28
- vLLM Semantic Router and AMD release Vela 3.0: 380M model beats GPT-5-mini at zero-shot — vllm_project · 2026-10-28
- User stuck with €104 of accidentally bought, unused ChatGPT credits and bot-only support — NickTsl · 2026-10-28
- Coding agent's compaction task fails with "model is at capacity" error — justalexoki · 2026-10-28