GPT-6.1 Sol replaces GPT-6 Sol in 7 days, scoring 1 point below Astra at quarter the cost
ArtificialAnlys · x · 2026-09-30
Artificial Analysis reviewed OpenAI's GPT-6.1 Sol, which replaced GPT-6 Sol just 7 days after its release. It scores 1 point below GPT-6 Astra on the Intelligence Index at $2/$10 per million tokens, with cache read discount raised from 90% to 95%.
Key takeaways
- +4 points vs GPT-6 Sol, +5 vs GPT-5.6 Sol on the Intelligence Index; Terminal-Bench 4.0 +12, HLE +5, hallucination rate down from 60% to 54%
- Cost efficiency: $0.72 per task at max effort — under a quarter of GPT-6 Astra ($3.26), 31% cheaper than GPT-6 Sol; every effort level pushes the cost Pareto frontier
- Uses 10–30% more output tokens than GPT-6 Sol, but low/medium efforts remain Pareto-optimal for token efficiency
- Coding Agent Index: +3 points over GPT-6 Sol at max effort, 2 points below Astra
Related event: GPT-6.1 Sol Matches Astra at a Fraction of the Cost Within a Week(4 posts)→
More from Models
- OpenAI DevDay: 20+ Updates Including Dots Agents, GPT-6.1 Sol, Ultrafast and $500 Pro 500 Plan — btibor91 · 2026-09-30
- Anthropic evals show GLM-5.3 achieves control-flow hijacks in 4% of trials, crossing a threshold — teortaxesTex · 2026-09-30
- Anecdotal Report: Claude Suddenly Stopped Getting Caught Cheating — sebpaquet · 2026-09-30
- Writing evals with per-sample rubrics may be exploitable via subtle benchmax overfitting — teortaxesTex · 2026-09-30
- GPT 6 lineage has a 'strange relationship with effort,' observer speculates — teortaxesTex · 2026-09-30
- Oxmiq Open-Sources NVFP4 Quantized Hy4 Model, Runs on RTX 6000 Pro — RajaXg · 2026-09-30