EnergyBench: GPT-6.2 Astra scores 87.9% at 3.6× cost, below GPT-6.1 Sol's 92.8%
emilahlback · x · 2026-11-18
Energy ran GPT-6.2 on EnergyBench, its enterprise workflow benchmark (6 cases × 3 trials, task score vs agent cost per task):
- GPT-6.1 Sol: 92.8% at $0.83/task
- GPT-6.2 Astra: 87.9% at $3.00/task (3.6× the cost)
- GPT-6.2 Sol: 81.7% at $0.97/task
Both GPT-6.2 variants underperform the previous-gen 6.1 Sol, with Astra costing far more; the team is sticking with 6.1 Sol. The chart also shows GPT-6.1 Terra at 62.7% and Luna Max at 43.0%.
More from Models
- Epoch AI benchmark: no LLM hits human employee standard, Fable 5.2 tops at 82% — Jsevillamol · 2026-11-18
- SpaceX CFO praises Grok Bot as 'probably the best agentic AI solution out there' — elonmusk · 2026-11-18
- UChicago study: Pangram AI-text detector beats GPTZero, Originality and RoBERTa with <5% error rates — _akpiper · 2026-11-18
- GPT-6.2 Astra hands-on: upgrades million-line legacy app, all tests green in 6 hours — thesaraharminta · 2026-11-18
- GPT-6.2 vs GPT-6 vs day-1 GPT-6: Astra image generation compared side by side — lyraxana · 2026-11-18
- Will Oremus article on negative parallelism fuels AI-writing detection debate — TuhinChakr · 2026-11-18