Skip to content

Benchmark

AHELM: MuTox - EM

Scores are percentages.

Models measured
12
Scores recorded
359
Sources
1
IQ weight
No weight published

Score distribution

75.5%
Mean
77.9%
Median
15.6%
Std dev
53.4%
10th percentile
100%
90th percentile
33.3%
Min
33.3%100%
The band covers the middle 80% of scores. The line marks the median.

Leaderboard

#ModelScore
201Gemini 2.5 Pro Preview 05-06language: Greek· AHELM (HELM Audio) data77%
202Gemini 2.0 Flashlanguage: Bengali· AHELM (HELM Audio) data76.5%
203Gemini 2.0 Flash Experimentallanguage: Bengali· AHELM (HELM Audio) data76.5%
204Gemini 2.5 Pro Preview 05-06language: Bengali· AHELM (HELM Audio) data76.5%
205Gemini 2.5 Flash Preview 05-20 (no think)language: Danish· AHELM (HELM Audio) data76%
206gemini-1.5-pro-001language: Bulgarian· AHELM (HELM Audio) data75.9%
207Gemini 2.5 Pro Preview 05-06language: Polish· AHELM (HELM Audio) data75.7%
208gemini-1.5-pro-002language: Czech· AHELM (HELM Audio) data75.2%
209Gemini 2.0 Flash Litelanguage: Italian· AHELM (HELM Audio) data75%
210gemini-1.5-flash-002language: Portuguese· AHELM (HELM Audio) data75%
211gemini-1.5-pro-001language: Portuguese· AHELM (HELM Audio) data75%
212gemini-1.5-pro-002language: Portuguese· AHELM (HELM Audio) data75%
213GPT-4o Audio Previewlanguage: Italian· AHELM (HELM Audio) data75%
214GPT-4o Audio Previewlanguage: Portuguese· AHELM (HELM Audio) data75%
215GPT-4o mini Audio Previewlanguage: Portuguese· AHELM (HELM Audio) data75%
216gemini-1.5-pro-001language: Danish· AHELM (HELM Audio) data74.9%
217gemini-1.5-flash-002· AHELM (HELM Audio) data73.7%
218Gemini 2.5 Pro Preview 05-06· AHELM (HELM Audio) data73.5%
219gemini-1.5-pro-002language: Hebrew· AHELM (HELM Audio) data73.4%
220Gemini 2.5 Flash Preview 05-20 (no think)language: Hungarian· AHELM (HELM Audio) data73.3%
221Gemini 2.5 Pro Preview 05-06language: Hungarian· AHELM (HELM Audio) data73.3%
222gemini-1.5-pro-001language: Finnish· AHELM (HELM Audio) data73%
223gemini-1.5-flash-002language: Greek· AHELM (HELM Audio) data72.3%
224Gemini 2.0 Flashlanguage: Urdu· AHELM (HELM Audio) data71.4%
225Gemini 2.0 Flash Experimentallanguage: Urdu· AHELM (HELM Audio) data71.4%
226Gemini 2.5 Flash Preview 05-20 (no think)language: Urdu· AHELM (HELM Audio) data71.4%
227Gemini 2.5 Flash Preview 05-20 (no think)language: Vietnamese· AHELM (HELM Audio) data71.4%
228Gemini 2.5 Pro Preview 05-06language: Urdu· AHELM (HELM Audio) data71.4%
229gemini-1.5-flash-001language: German· AHELM (HELM Audio) data71.4%
230gemini-1.5-flash-001language: Hindi· AHELM (HELM Audio) data71.4%
231gemini-1.5-flash-002language: German· AHELM (HELM Audio) data71.4%
232gemini-1.5-flash-002language: Vietnamese· AHELM (HELM Audio) data71.4%
233gemini-1.5-pro-001language: Hindi· AHELM (HELM Audio) data71.4%
234gemini-1.5-pro-001language: Vietnamese· AHELM (HELM Audio) data71.4%
235gemini-1.5-pro-002language: Vietnamese· AHELM (HELM Audio) data71.4%
236Qwen2.5-Omni 7Blanguage: Urdu· AHELM (HELM Audio) data71.4%
237Qwen2.5-Omni 7Blanguage: Vietnamese· AHELM (HELM Audio) data71.4%
238gemini-1.5-pro-001language: Polish· AHELM (HELM Audio) data70.4%
239gemini-1.5-pro-002language: Hungarian· AHELM (HELM Audio) data70.3%
240Gemini 2.0 Flashlanguage: Arabic· AHELM (HELM Audio) data70%
241Gemini 2.0 Flash Experimentallanguage: Arabic· AHELM (HELM Audio) data70%
242Gemini 2.0 Flash Litelanguage: Arabic· AHELM (HELM Audio) data70%
243gemini-1.5-flash-002language: Swahili· AHELM (HELM Audio) data70%
244GPT-4o Audio Previewlanguage: Arabic· AHELM (HELM Audio) data70%
245gemini-1.5-flash-002language: Slovak· AHELM (HELM Audio) data69.9%
246GPT-4o Audio Previewlanguage: Spanish· AHELM (HELM Audio) data69.4%
247gemini-1.5-pro-001language: Slovak· AHELM (HELM Audio) data69.4%
248gemini-1.5-flash-002language: Estonian· AHELM (HELM Audio) data69.3%
249Gemini 2.5 Pro Preview 05-06language: Dutch· AHELM (HELM Audio) data69.2%
250gemini-1.5-flash-001language: Dutch· AHELM (HELM Audio) data69.2%

Showing 201–250 of 359 results.