HATEBENCH
● Models vs Hateful Hindi Text
ACCURACY DISTRIBUTION
SUCCESS RATE ON HINDI HATE SPEECH DEFINITIONS
01
gemini-3.1-flash-lite
75%
02
trinity-large-preview
71%
03
gpt-5.2-pro
71%
04
hy3-preview
71%
05
gemini-3-flash-preview
70%
06
gemini-3.1-flash-lite-preview
70%
07
gemma-4-31b-it
70%
08
gemma-3n-e4b-it
70%
09
claude-opus-4.7
67%
10
claude-opus-4.6
67%
11
claude-sonnet-4.6
67%
12
gpt-5.2
67%
13
gpt-5.5-pro
67%
14
deepseek-v4-flash
67%
15
gemini-3-pro-preview
67%
16
gemma-3-12b-it
67%
17
gpt-5.4-pro
67%
18
qwen3.5-plus-02-15
67%
19
gemini-3.5-flash
66%
20
gpt-5.3-chat
66%
21
kimi-k2.5
66%
22
grok-4.1-fast
66%
23
gemma-3-27b-it
65%
24
qwen3.7-max
65%
25
glm-5-turbo
65%
26
grok-4.3
65%
27
nova-2-lite-v1
65%
28
qwen3.6-max-preview
65%
29
claude-sonnet-4
64%
30
gpt-5.4
64%
31
gpt-5.4-nano
64%
32
command-a
63%
33
qwen3.5-plus-20260420
63%
34
qwen3.6-flash
63%
35
glm-5.1
63%
36
glm-4.7
63%
37
seed-2.0-lite
63%
38
mimo-v2-pro
63%
39
gpt-5.4-mini
62%
40
gpt-5.5
62%
41
sarvam-105b
61%
42
deepseek-v4-pro
61%
43
glm-5
61%
44
step-3.5-flash
61%
45
mimo-v2-omni
60%
46
deepseek-chat-v3.1
59%
47
minimax-m2.1
59%
48
sarvam-30b
59%
49
qwen3-32b
59%
50
ministral-14b-2512
59%
51
deepseek-v3.2
58%
52
llama-3.3-nemotron-super-49b-v1.5
58%
53
qwen3-235b-a22b-2507
57%
54
mistral-medium-3-5
56%
55
mistral-small-2603
55%
56
qwen3.5-9b
54%
57
minimax-m2.7
54%
58
minimax-m2.5
53%
59
llama-4-maverick
51%
60
hunyuan-a13b-instruct
51%
61
mercury-2
49%
62
phi-4
47%
63
molmo-2-8b
43%