Research made with GPT-6.1 Sol

GPT-6.1 Sol replaces GPT-6 Sol after just 7 days. It scores 1 point below GPT-6 Astra in the Intelligence Index at less than one quarter of the Cost per Task

Pricing matches GPT-6 Sol at $2/$10 per million input/output tokens, except that the cache read discount rises from 90%…

🚨 GPT-6.1 Sol has a serious Big Problem 😭

GPT-6.1 Sol is too slow I've been Testing it for the last 24h

6.1 Sol xHigh: ~20.7 tok/s
6 Sol xHigh: ~53.8 tok/s
5.6 Sol xHigh: ~46.7 tok/s
6 Sol Max: ~51.7 tok/s

That’s a massive slowdown compared to the previous Sol models…

I gave the same task to GPT-6 Astra, GPT-6.1 Sol and Grok 4.7 (all with xhigh reasoning) and here is the result.

Overall score:

• Opus 5.5: 9.1
• Astra: 8.0
• Grok 4.7 and Sol 6.1: 7.7
• Sol 5.6: 7.5
• Sol 6: 6.9

Opus 5.5 is still a clear winner.

17 cents of GPT-6.1 Sol made a local RTX 3090 + Qwen3.8-27B dramatically faster at coding.

That works until you look at the third comparison.

They had the models build the same 3 small games:
🏠 Qwen alone: 114.2 min / $0 cloud
🤝 Sol plans + Qwen codes: 43.4 min / $0.17
☁️…

Every GPT-6.1 Sol effort setting on our internal writing benchmark (181 model configs in total):

xhigh: #53, 1976 Elo, $0.13 and 5.8 min per script
max: #55, 1974 Elo, $0.18, 8.0 min
low: #60, 1950 Elo, $0.05, 1.5 min
medium: #64, 1926 Elo, $0.05, 1.4 min
high: #67, 1910 Elo…

Short proofs of Graffiti conjectures 198, 219, 252, 254, 584 and 712 (S. Fajtlowicz, Written on the Wall) using Claude Opus 5.5 and GPT-6.1 Sol

A proof of Graffiti conjecture 262 (S. Fajtlowicz, Written on the Wall) using Claude Opus 5.5 and GPT-6.1 Sol

Захват территории: турнир команд, которые целиком создают нейросети (Claude Sonnet 5.5 vs GPT-6.1 Sol)

Claude Sonnet 5.5 vs GPT-6.1 Sol: four jobs, three attempts each. Briefs, checks and every attempt, so you can test the apps yourself.

GPT-6.1 Sol benchmark data (2026-09-30): 7 tracks, 3,527 calls vs GPT-6 Sol/Astra/Luna & GPT-5.6 Sol — scorecard, per-item results, task sets (data only, CC BY 4.0)