Multiple leaks claim Claude Sonnet 5.5 decisively outscores GPT-6 sol

On September 29, multiple third-party sources leaked that Anthropic's Claude Sonnet 5.5 decisively outscores OpenAI's GPT-6 sol on benchmarks and approaches — or in some claims surpasses — GPT-6 Astra, sparking debate over OpenAI's competitiveness. None of the figures have been officially confirmed.

Confirmed

X user ns123abc (84k followers) posted screenshots claiming GPT-6 sol was "brutally mogged" by Sonnet 5.5.
LuminaBench leaked that Sonnet 5.5 scored 56 on the Artificial Analysis Intelligence Index versus 48 for Sol, and also highlighted strong Terminal-Bench results.
Another leaker said Sonnet 5.5's score of 56 ranks just below Claude Opus 5.5 and above GPT-6 Astra.
Account AIScreening posted third-party test screenshots showing Sonnet 5.5 far ahead of GPT-6 sol and close to GPT-6 Astra.
A cited post said Claude Sonnet 5.5 Max ranks second on the AA Intelligence Index, above GPT-6 Astra Max and just below Opus-5.5.

Unconfirmed

All figures come from leaks and third-party reposts; benchmarks, prompts and version details are unclear and unverified by Anthropic, OpenAI or Artificial Analysis.
The leaks contradict each other: LuminaBench implies Sonnet 5.5 (56) surpasses Opus 5.5, while another says it ranks just below Opus 5.5.

Why it matters

If true, Anthropic's mid-tier Sonnet 5.5 would outmuscle OpenAI's GPT-6 sol and rival GPT-6 Astra, reshaping user choices and the market narrative between the two vendors.
Multiple independent accounts spreading similar conclusions the same day signals rising community concern over OpenAI's slipping frontier-model standing, even if the exact numbers are imprecise.

2026-09-29 ~ 2026-09-29 · 5 related posts

Full story(10 episodes)→

Episode 1: Insiders claim OpenAI and Anthropic hold models far ahead of public releases(2026-09-24, 2 posts)
Episode 2: Anthropic Reportedly Set to Ship Sonnet/Haiku 5.5 Ahead of OpenAI DevDay(2026-09-26, 2 posts)
Episode 3: Claude Sonnet 5.5 leak: gradual rollout underway, launch imminent(2026-09-27, 20 posts)
Episode 4: Insider Claims OpenAI Is Sitting on a Stronger Model(2026-09-28, 2 posts)
Episode 5: Rumors: Anthropic Ships Opus 5.5 and Sonnet 5.5 Ahead of OpenAI DevDay(2026-09-29, 7 posts)
Episode 6: Anthropic Launches Claude Sonnet 5.5: Over 30% Faster, Up to 30% Cheaper(2026-09-29, 58 posts)
Episode 7: Claude Sonnet 5.5 launches near-flagship intelligence amid token-cost controversy(2026-09-29, 27 posts)
Episode 8: Multiple leaks claim Claude Sonnet 5.5 decisively outscores GPT-6 sol(2026-09-29, 5 posts)
Episode 9: Data comparison sparks debate: Sonnet 5.5 called a flop, GPT-6 Sol wins on value(2026-09-29, 2 posts)
Episode 10: Sonnet 5.5 Nearly Matches Flagship Opus in Real-World Use(2026-09-29, 3 posts)

Primary sources

Leak claims OpenAI GPT-6 sol brutally outclassed by Claude Sonnet 5.5 — ns123abc · 2026-09-29 02:44 (UTC+08:00)
Leaked benchmarks claim Claude Sonnet 5.5 hits 56 on AA index, beating Opus 5.5 — airesearch12 · 2026-09-29 03:46 (UTC+08:00)
Claude Sonnet 5.5 Max Reportedly Ranks Second on AI Index, Above GPT-6, Near Opus-5.5 — airesearch12 · 2026-09-29 03:49 (UTC+08:00)

Claude Sonnet 5.5 Reportedly Scores 56 on Artificial Analysis Intelligence Index — thesaraharminta · 2026-09-29
[source] Leak claims OpenAI GPT-6 sol brutally outclassed by Claude Sonnet 5.5 — ns123abc · 2026-09-29
[source] Leaked benchmarks claim Claude Sonnet 5.5 hits 56 on AA index, beating Opus 5.5 — airesearch12 · 2026-09-29
[source] Claude Sonnet 5.5 Max Reportedly Ranks Second on AI Index, Above GPT-6, Near Opus-5.5 — airesearch12 · 2026-09-29
Third-party benchmark shows Claude Sonnet 5.5 crushing GPT-6 sol, nearly matching GPT-6 Astra — rickasaurus · 2026-09-29