Benchmark
Public rankings, methodology, reports, and the African AI Trust Index.
benchmark.mlatho.comWhat the Index scores
Five headline scores journalists and product teams can cite — produced from a fixed, human-judged evaluation set.
01
African Language Score
Comprehension, instruction-following, and translation quality on African-language prompts (Preview focus: Chichewa ↔ English).
02
Cultural Understanding Score
Checkable local geography, institutions, and norms — where fluent English often hides confident error.
03
Business Readiness Score
Practical usefulness for African service and commerce contexts (channels, local constraints, usable answers).
04
Safety Score
Africa-relevant refusal and caution: scams, medical overclaim, election-adjacent heat, and demeaning content.
05
Overall Trust Score
Weighted composite across language, culture, business readiness, and safety for this edition’s fixed set.
Latest rankings
According to the Mlatho African AI Trust Index — models are scored under a frozen prompt template, without tools, by bilingual human raters.
Loading leaderboard…
Fetching the latest published scores.
The African AI Trust Index Preview is a small human-evaluated set focused on Chichewa and African context. Scores are for this set only, under a fixed prompt template without tools. They are not a universal ranking of model quality.
How the ecosystem fits
Lab invents. Benchmark proves. App monetizes. Learn trains the people who make it possible.
App · Evaluations
Commercial product: submit models, run jobs, download reports, manage projects.
app.mlatho.comLab
Research engine for new benchmarks, datasets, and experimental metrics.
lab.mlatho.comLearn
Trains the African evaluation workforce that produces gold and adjudication.
learn.mlatho.comNeed a private run on your model?
Product teams shipping to African users can commission a scoped private evaluation against an expanded set — same method discipline, confidential outputs.