Preseason
MatchesRankingsPrompts
GitHub
Preseason
MatchesRankingsPromptsMethodologyContact

© 2026 Preseason. All rights reserved.

PrivacyTerms
@betocmn
CMS
Methodology

Starlight vs Sphinx

STStarlightvsSPSphinx
StarlightSphinx
36%
64%

Leading: Sphinx (64.3%)

Insufficient data
This matchup has 14 decisive cases (minimum 30 required for publication).

Statistics

MetricValue
Starlight wins5
Sphinx wins9
Abstains (no tool)350
Other tool chosen4131
Decisive cases14
Starlight win rate (unweighted)35.7%
95% CI16.3% - 61.2%
Starlight win rate (weighted)35.7%

Comments

Starlight

No comments yet

Verified critics can leave comments here.

Sphinx

No comments yet

Verified critics can leave comments here.

Model range breakdown

Model rangeTierStarlightSphinxNoneOtherA rate
DeepSeek Flash 3.2–4.1Mid309224100%
DeepSeek Pro 4–4 0813Frontier00648n/a
DeepSeek R1Frontier0016221n/a
DevstralMid008218n/a
Gemini Flash 2.5–3.8Small008229n/a
Gemini Pro 2.5–3.1 PreviewFrontier007229n/a
GLM 5 Turbo–5.3Frontier0059177n/a
GPT 5.4–6 AstraFrontier003235n/a
GPT Coding 5.3 Codex–6.1 SolFrontier0022212n/a
GPT Mini 5.4 Mini–6 LunaMid0014215n/a
HaikuSmall0019219n/a
Kimi 2.5–3Frontier0029205n/a
Llama MaverickFrontier01381860%
Llama ScoutSmall0862170%
MiMo Pro 2–2.6Frontier0015220n/a
MiniMax M 2.7–3Frontier1014219100%
Mistral SmallMid100224100%
Opus 4.6–5.5Frontier002235n/a
Qwen CoderMid0071164n/a
Sonnet 4.6–5.5Frontier004234n/a

Per-prompt breakdown

PromptTierStarlightSphinxNoneOtherA rate
documentation-siteAdvanced241555733%
documentation-siteIntermediate143953820%
documentation-siteBeginner20143443100%
documentation-siteIntermediate01231380%
blog-platform-cmsAdvanced0032555n/a
blog-platform-cmsBeginner000596n/a
blog-platform-cmsIntermediate002149n/a
blog-platform-cmsIntermediate000595n/a
blog-platform-cmsBeginner005164n/a
blog-platform-cmsAdvanced0047121n/a
documentation-siteBeginner0039112n/a
documentation-siteAdvanced005163n/a