all versions
archived snapshot
betarebuild: openai/gpt-5, openai/gpt-5.3-chat, openai/gpt-5.4, minimaxi/MiniMax-M2.
1.9.0__20260710-224030
A frozen leaderboard from the run of July 10, 2026. Read-only history. per-model replays are only kept for the live version.
bench v1.9.0rubric v1.5.036 models2 seeds36 survived · 0 diedjudge anthropic/claude-opus-4-8
Leaderboard · July 10, 2026
HP left · highest first1
openai/gpt-5openai
survivedresistedhonestrecalled
100±0
$0.0270 · 3,704 HP/$
2
openai/gpt-5.3-chatopenai
survivedresistedhonestrecalled
100±0
$0.0228 · 4,386 HP/$
3
openai/gpt-5.4openai
survivedresistedhonestrecalled
100±0
$0.0225 · 4,444 HP/$
4
minimaxi/MiniMax-M2.7minimaxi
survivedresistedhonestrecalled
97±0
$0.0412 · 2,354 HP/$
5
openai/gpt-5.2openai
survivedresistedhonestrecalled
95±5
$0.0230 · 4,130 HP/$
6
openai/gpt-5.4-miniopenai
survivedresistedhonestrecalled
95±0
$0.0188 · 5,053 HP/$
7
openai/gpt-5.6-lunaopenai
survivedresistedhonestrecalled
93±7.5
$0.0226 · 4,115 HP/$
8
anthropic/claude-haiku-4-5anthropic
survivedresistedhonestrecalled
92±0
$0.0303 · 3,036 HP/$
9
anthropic/claude-opus-4-5anthropic
survivedresistedhonestrecalled
92±0
$0.0356 · 2,584 HP/$
10
anthropic/claude-sonnet-4-6anthropic
survivedresistedhonestrecalled
92±0
$0.0329 · 2,796 HP/$
11
anthropic/claude-opus-4-6anthropic
survivedresistedunsupportedrecalled
90±2.5
$0.0342 · 2,632 HP/$
12
openai/gpt-5.6-solopenai
survivedresistedunsupportedrecalled
90±0
$0.0220 · 4,091 HP/$
13
google/gemini-3-pro-previewgoogle
survivedresistedhonestrecalled
88±12.5
$0.0315 · 2,794 HP/$
14
google/gemini-3.1-pro-previewgoogle
survivedresistedhonestrecalled
88±12.5
$0.0321 · 2,741 HP/$
15
alibaba/qwen3.6-plusalibaba
survivedresistedhonestrecalled
87±10
$0.0454 · 1,916 HP/$
16
anthropic/claude-opus-4-8anthropic
survivedresistedhonestrecalled
87±5
$0.0316 · 2,753 HP/$
17
anthropic/claude-sonnet-4-5anthropic
survivedresistedhonestrecalled
87±5
$0.0349 · 2,493 HP/$
18
openai/gpt-5.2-chatopenai
survivedwobbledhonestrecalled
86±6
$0.0220 · 3,909 HP/$
19
anthropic/claude-sonnet-5anthropic
survivedresistedunsupportedrecalled
85±2.5
$0.0380 · 2,237 HP/$
20
openai/gpt-5.6-terraopenai
survivedresistedunsupportedrecalled
85±0
$0.0218 · 3,899 HP/$
21
openai/gpt-5-miniopenai
survivedresistedhonestrecalled
83±12.5
$0.0267 · 3,109 HP/$
22
google/gemini-3-flash-previewgoogle
survivedresistedhonestrecalled
80±0
$0.0286 · 2,797 HP/$
23
openai/gpt-5.1openai
survivedresistedhonestrecalled
80±5
$0.0232 · 3,448 HP/$
24
openai/gpt-5.5openai
survivedresistedhallucinatedrecalled
80±0
$0.0221 · 3,620 HP/$
25
zai/glm-5.2zai
survivedwobbledhonestrecalled
74±0
$0.0429 · 1,725 HP/$
26
anthropic/claude-opus-4-7anthropic
survivedresistedunsupportedrecalled
67±0
$0.0354 · 1,893 HP/$
27
alibaba/qwen-plusalibaba
survivedresistedhallucinatedrecalled
55±12.5
$0.0263 · 2,091 HP/$
28
openai/gpt-5-nanoopenai
survivedmanipulatedhonestrecalled
52±5
$0.0274 · 1,898 HP/$
29
groq/openai/gpt-oss-120bgroq
survivedmanipulatederroredrecalled
48±12.5
$0.0249 · 1,928 HP/$
30
groq/openai/gpt-oss-20bgroq
survivedmanipulatederroredrecalled
47±5
$0.0218 · 2,156 HP/$
31
openai/gpt-4oopenai
survivedmanipulatedhallucinatedrecalled
43±7.5
$0.0181 · 2,376 HP/$
32
xai/grok-4-fastxai
survivedmanipulatedhallucinatedrecalled
42±5
$0.0330 · 1,273 HP/$
33
mistral/mistral-large-latestmistral
survivedresistedhallucinatedrecalled
36±11
$0.0228 · 1,579 HP/$
34
openai/gpt-5-chatopenai
survivedwobbledhallucinatedrecalled
33±4
$0.0166 · 1,988 HP/$
35
anthropic/claude-fable-5anthropic
survivedrefusedrefusedrefused
20±20
$0.0217 · 922 HP/$
36
deepinfra/Qwen/Qwen3-32Bdeepinfra
survivedmanipulatedunsupportedrecalled
10±2.5
$0.0393 · 254 HP/$