‹ PublicAI Index
The LLM benchmark aggregator.
Aya 23 35B
Cohere · 35B
Strongest in Jailbreak resistance (#99 of 272), weakest in Toxicity avoidance (#249 of 272). Above par in 1 of 6 scopes. Among the models it meets almost everywhere, it finishes behind Claude Sonnet 5 and Claude Opus 5.5 and ahead of Gemma 3 4B It and Mistral Small 24B Instruct.
§ 1 · Profile
What it is good at
Bars run from 50 — the average of the models each source lists — so right of the line is above par. The middle column is the gap to whoever leads that scope.
Safety47.4−14.1#242/3361/3
Jailbreak resistance56.3−8.6#99/2721/1
Fairness46.5−23.5#159/2991/2
Secure code48.6−17.9#173/2741/1
Harm refusal48.3−13.3#184/2991/2
Toxicity avoidance35−24#249/2721/1
§ 2 · Head to head
What it beats, and what beats it
The same models turn up scope after scope. Counted once: where both were placed, who finished higher. Bars run right for Aya 23 35B, left for the other.
§ 3 · Sources
Where the numbers come from
1 publication, 6 figures. Every one links to the page it was read from.
Enkrypt · Jailbreak risk 7.5%Enkrypt · Harmful content risk 53.9%Enkrypt · CBRN risk 8.3%Enkrypt · Toxicity risk 16.8%Enkrypt · Bias risk 83.2%Enkrypt · Insecure code risk 36.4%
Badge
[](https://publicai.io/model-index/m/aya-23-35b)