AfriHate
Models classify hate speech and abusive language across African languages.
AfroBench lists F1 as the metric for hate speech detection.
- Made by
- Shamsuddeen Hassan Muhammad and colleagues
- Released
- January 2025
- Kind
- Reading in many languages
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Multiple choice
- Tasks
- All public
- Measures
- Capability
- Risk
- Hate
- Language
- Many languages
- Interaction
- Single turn
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Many languages, Content safety
- Strongest model tested
-
Llama 3.1-70B #177 on the capability index
- Last model added
- July 2024
- Scores refreshed
- November 20th, 2026
AfroXLMR-76L tops the leaderboard with 78.2% Gemma 2 27B leads with 58.8% among models released by July 2024
Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is Gemma 2 27B, at 58.8%. This view shows only models we track, each at its best.
The newest model with a score here came out in July 2024, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 1 AfroXLMR-76L Multilingual fine-tuning 78.2%
- 2 AfroXLMR Monolingual fine-tuning 76.2%
- 3 AfriBERTa Monolingual fine-tuning 72.3%
- 4 GPT-4o (snapshot unspecified) 20-shot 71.7%
- 5 AfriTeVa V2 Monolingual fine-tuning 68.7%
- 6
Gemma 2 27B 20-shot 58.8%
- 7
Llama 3.1-70B 20-shot 57.1%
- 8
Gemma 2 9B 20-shot 56.3%
- 9 LaBSE 20-shot 53%
- 10 aya-23-35B 20-shot 48.8%
- 11
Llama 3.1-8B 20-shot 42.7%
- 12
Mistral 7B v0.1 20-shot 41.4%
- 13 InkubaLM-0.4B 10-shot 25.4%
- 14 bloomz-7b1-mt 20-shot 25.4%
- 15 mt0-small 5-shot 22%
Source: The AfriHate paper
- 1
Gemma 2 27B 20-shot 58.8%
- 2
Llama 3.1-70B 20-shot 57.1%
- 3
Gemma 2 9B 20-shot 56.3%
- 4
Llama 3.1-8B 20-shot 42.7%
- 5
Mistral 7B v0.1 20-shot 41.4%
Source: The AfriHate paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | AfroXLMR-76L | 78.2% | Multilingual fine-tuning | Official leaderboard | |
| 2 | AfroXLMR | 76.2% | Monolingual fine-tuning | Official leaderboard | |
| 3 | AfriBERTa | 72.3% | Monolingual fine-tuning | Official leaderboard | |
| 4 | GPT-4o (snapshot unspecified) | 71.7% | 20-shot | Official leaderboard | |
| 5 | AfriTeVa V2 | 68.7% | Monolingual fine-tuning | Official leaderboard | |
| 6 | | 58.8% | 20-shot | Jun 2024 | Official leaderboard |
| 7 | | 57.1% | 20-shot | Jul 2024 | Official leaderboard |
| 8 | | 56.3% | 20-shot | Jun 2024 | Official leaderboard |
| 9 | LaBSE | 53% | 20-shot, SetFit | Official leaderboard | |
| 10 | aya-23-35B | 48.8% | 20-shot | Official leaderboard | |
| 11 | | 42.7% | 20-shot | Jul 2024 | Official leaderboard |
| 12 | | 41.4% | 20-shot | Sep 2023 | Official leaderboard |
| 13 | InkubaLM-0.4B | 25.4% | 10-shot | Official leaderboard | |
| 14 | bloomz-7b1-mt | 25.4% | 20-shot | Official leaderboard | |
| 15 | mt0-small | 22% | 5-shot | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 58.8% | 20-shot | Jun 2024 | Official leaderboard |
| 2 | | 57.1% | 20-shot | Jul 2024 | Official leaderboard |
| 3 | | 56.3% | 20-shot | Jun 2024 | Official leaderboard |
| 4 | | 42.7% | 20-shot | Jul 2024 | Official leaderboard |
| 5 | | 41.4% | 20-shot | Sep 2023 | Official leaderboard |
Rankings by part
We read 15 parts from the benchmark’s paper, each ranking models on one part of the test, such as amh, ary and arq. Gemma 2 27B leads 10 of them, and Llama 3.1-70B leads the other 5. Arrows show how far a model moved from its overall rank among the models we track.
amh
5 models we track have a score here.
- 1
Gemma 2 27B 62.3%
- 2
Llama 3.1-70B 61.2%
- 3
Gemma 2 9B 61%
- 4
Llama 3.1-8B 47.7%
- 5
Mistral 7B v0.1 39.7%
ary
5 models we track have a score here.
- 1
Llama 3.1-70B ↑1 66.2%
- 2
Gemma 2 27B ↓1 64.1%
- 3
Gemma 2 9B 61.1%
- 4
Llama 3.1-8B 43.1%
- 5
Mistral 7B v0.1 36.5%
arq
5 models we track have a score here.
- 1
Gemma 2 27B 65.8%
- 2
Gemma 2 9B ↑1 64.4%
- 3
Llama 3.1-70B ↓1 63.8%
- 4
Llama 3.1-8B 57.9%
- 5
Mistral 7B v0.1 52.9%
hau
5 models we track have a score here.
- 1
Gemma 2 27B 59.6%
- 2
Llama 3.1-70B 57.4%
- 3
Gemma 2 9B 56.6%
- 4
Llama 3.1-8B 41.6%
- 5
Mistral 7B v0.1 38.4%
ibo
5 models we track have a score here.
- 1
Gemma 2 27B 58.8%
- 2
Llama 3.1-70B 55.1%
- 3
Gemma 2 9B 54.6%
- 4
Mistral 7B v0.1 ↑1 39%
- 5
Llama 3.1-8B ↓1 37.8%
kin
5 models we track have a score here.
- 1
Gemma 2 27B 66.1%
- 2
Gemma 2 9B ↑1 64.1%
- 3
Llama 3.1-70B ↓1 62.5%
- 4
Mistral 7B v0.1 ↑1 52.3%
- 5
Llama 3.1-8B ↓1 48.9%
oro
5 models we track have a score here.
- 1
Llama 3.1-70B ↑1 49%
- 2
Gemma 2 27B ↓1 48.9%
- 3
Gemma 2 9B 45.8%
- 4
Llama 3.1-8B 39.8%
- 5
Mistral 7B v0.1 39.5%
pcm
5 models we track have a score here.
- 1
Llama 3.1-70B ↑1 64.4%
- 2
Gemma 2 27B ↓1 63.2%
- 3
Gemma 2 9B 62.5%
- 4
Llama 3.1-8B 55.4%
- 5
Mistral 7B v0.1 54.6%
som
5 models we track have a score here.
- 1
Llama 3.1-70B ↑1 54.4%
- 2
Gemma 2 27B ↓1 54.1%
- 3
Gemma 2 9B 52.2%
- 4
Mistral 7B v0.1 ↑1 34.1%
- 5
Llama 3.1-8B ↓1 33%
swa
5 models we track have a score here.
- 1
Gemma 2 27B 81.2%
- 2
Llama 3.1-70B 77.7%
- 3
Gemma 2 9B 77.6%
- 4
Llama 3.1-8B 69.4%
- 5
Mistral 7B v0.1 56%
tir
5 models we track have a score here.
- 1
Llama 3.1-70B ↑1 53.9%
- 2
Gemma 2 27B ↓1 53.4%
- 3
Gemma 2 9B 48.4%
- 4
Mistral 7B v0.1 ↑1 37.4%
- 5
Llama 3.1-8B ↓1 37.1%
twi
5 models we track have a score here.
- 1
Gemma 2 27B 50.9%
- 2
Gemma 2 9B ↑1 48.9%
- 3
Llama 3.1-70B ↓1 47.5%
- 4
Llama 3.1-8B 34.8%
- 5
Mistral 7B v0.1 32%
xho
5 models we track have a score here.
- 1
Gemma 2 27B 43.1%
- 2
Llama 3.1-70B 40.3%
- 3
Gemma 2 9B 39%
- 4
Mistral 7B v0.1 ↑1 34.2%
- 5
Llama 3.1-8B ↓1 29.2%
yor
5 models we track have a score here.
- 1
Gemma 2 27B 61.7%
- 2
Gemma 2 9B ↑1 60.5%
- 3
Llama 3.1-70B ↓1 58.1%
- 4
Mistral 7B v0.1 ↑1 46.4%
- 5
Llama 3.1-8B ↓1 42.7%
zul
5 models we track have a score here.
- 1
Gemma 2 27B 56.9%
- 2
Llama 3.1-70B 56.4%
- 3
Gemma 2 9B 53.7%
- 4
Mistral 7B v0.1 ↑1 35%
- 5
Llama 3.1-8B ↓1 30.9%
Scores at each effort level
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: The AfriHate paper
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| 20-shot | 58.8% | ||||
| 10-shot | 57.9% | ||||
| 5-shot | 57.2% | ||||
| 20-shot | 57.1% | ||||
| 10-shot | 56.5% | ||||
| 20-shot | 56.3% | ||||
| 10-shot | 55.6% | ||||
| 5-shot | 55.2% | ||||
| 5-shot | 52.8% | ||||
| 20-shot | 42.7% | ||||
| 20-shot | 41.4% | ||||
| 10-shot | 40.7% | ||||
| 10-shot | 39.9% | ||||
| 5-shot | 38.8% | ||||
| 5-shot | 38.7% | ||||
| 0-shot | 36.6% | ||||
| 0-shot | 36.5% | ||||
| 0-shot | 21.3% | ||||
| 0-shot | 19.3% | ||||
| 0-shot | 12.5% |
Scores at each effort level
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: The AfriHate paper
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| 20-shot | 58.8% | ||||
| 10-shot | 57.9% | ||||
| 5-shot | 57.2% | ||||
| 20-shot | 57.1% | ||||
| 10-shot | 56.5% | ||||
| 20-shot | 56.3% | ||||
| 10-shot | 55.6% | ||||
| 5-shot | 55.2% | ||||
| 5-shot | 52.8% | ||||
| 20-shot | 42.7% | ||||
| 20-shot | 41.4% | ||||
| 10-shot | 40.7% | ||||
| 10-shot | 39.9% | ||||
| 5-shot | 38.8% | ||||
| 5-shot | 38.7% | ||||
| 0-shot | 36.6% | ||||
| 0-shot | 36.5% | ||||
| 0-shot | 21.3% | ||||
| 0-shot | 19.3% | ||||
| 0-shot | 12.5% |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The AfriHate paper
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official AfriHate leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Multilingual
- MMMLU In use MMLU professionally translated into 14 languages. Multiple-choice accuracy. Claude Mythos Preview 92.7%
- WMT24++ In use English-to-55-language translation of WMT24 test data. Scored with automatic metrics such as MetricX or xCOMET.
K-EXAONE 90.5%
- MMLU-ProX In use MMLU-Pro translated into up to 29 languages with expert review. Multiple-choice accuracy per language.
Qwen3.7-Max 87%
- CLIcK In use Models answer 1,995 multiple-choice questions on Korean language and culture. Questions come from official Korean exams and textbooks in eleven categories.
Solar Open 2 90.7%
- Global MMLU Lite 3.0 In use Models answer 400 human-translated MMLU questions per language in 23 languages. Version 3.0 adds Czech, Hungarian, Oriya, Slovak and Tajik and updates Italian. Accuracy. Claude Opus 5 95.7%
- MMMB In use Models answer multiple-choice questions about images in six languages across 15 categories. The set has about 12,000 questions. LFM2.5-VL-3B 83%
Data audit
9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 2 years ago, in July 2024, for Llama 3.1-70B. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
- Pass Independent runs Scores from the benchmark’s paper (15 entries).
- Pass Models scored 5 models we track, plus 10 other entries on its leaderboard.
- Warning Newest score The newest came 2 years ago, in July 2024, for Llama 3.1-70B.
- Note Frontier models None of the top 10 has a score. The strongest tested is Llama 3.1-70B (#177).
- Pass Headroom The best is 58.8%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on arxiv.org: 20 rows for models we track and 10 other entries at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.