# BullshitBench v2

> A benchmark that tests whether AI models challenge nonsensical, ill-posed, or logically flawed prompts instead of confidently generating incorrect answers. Measures the critical ability to push back on bad input.

Canonical page: https://benchlm.ai/benchmarks/bullshitbenchv2

- Category: [Reasoning](/reasoning)
- Last updated: September 10, 2026 at 8:20 PM UTC

## About BullshitBench v2

- Year: 2025
- Tasks: Nonsensical and flawed prompts across multiple domains
- Format: Prompt challenge and refusal evaluation
- Difficulty: Robustness and critical reasoning
- Paper: [BullshitBench: Measuring whether AI models challenge nonsensical prompts](https://petergpt.github.io/bullshit-benchmark/)

BullshitBench evaluates a crucial real-world capability: knowing when NOT to answer. Models that score highly recognize flawed premises, impossible physics scenarios, and logical contradictions rather than hallucinating plausible-sounding responses. V2 includes harder and more diverse challenge categories.

BullshitBench v2 is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (214 models)

| Rank | Model | Creator | Score |
|------|-------|---------|-------|
| 1 | [Claude Opus 4.8](/models/claude-opus-4-8) | Anthropic | 95% |
| 2 | [Qwen3.8 Max](/models/qwen3-8-max) | Alibaba | 95% |
| 3 | [Claude Opus 4.8](/models/claude-opus-4-8) | Anthropic | 94% |
| 4 | [Claude Sonnet 4.6](/models/claude-sonnet-4-6) | Anthropic | 91% |
| 5 | [Claude Opus 4.5](/models/claude-opus-4-5) | Anthropic | 90% |
| 6 | [Claude Sonnet 4.6](/models/claude-sonnet-4-6) | Anthropic | 89% |
| 7 | [Claude Opus 4.6](/models/claude-opus-4-6) | Anthropic | 87% |
| 8 | [Claude Opus 4.6](/models/claude-opus-4-6) | Anthropic | 83% |
| 9 | [Claude Opus 4.7](/models/claude-opus-4-7) | Anthropic | 83% |
| 10 | [Claude Sonnet 5](/models/claude-sonnet-5) | Anthropic | 80% |
| 11 | [Claude Sonnet 4.5](/models/claude-sonnet-4-5) | Anthropic | 79% |
| 12 | [Claude Opus 4.5](/models/claude-opus-4-5) | Anthropic | 79% |
| 13 | [Qwen3.8-27B](/models/qwen3-8-27b) | Alibaba | 79% |
| 14 | [Claude Sonnet 5](/models/claude-sonnet-5) | Anthropic | 78% |
| 15 | [Qwen3.5 397B (Reasoning)](/models/qwen3-5-397b-reasoning) | Alibaba | 78% |
| 16 | [Claude Fable 5.1](/models/claude-fable-5-1) | Anthropic | 77% |
| 17 | [Claude Haiku 4.5](/models/claude-haiku-4-5) | Anthropic | 77% |
| 18 | [Kimi K3](/models/kimi-k3) | Moonshot AI | 74% |
| 19 | [Claude Opus 4.7](/models/claude-opus-4-7) | Anthropic | 74% |
| 20 | [Claude Sonnet 4.5](/models/claude-sonnet-4-5) | Anthropic | 74% |
| 21 | [Claude Opus 5](/models/claude-opus-5) | Anthropic | 73% |
| 22 | [Kimi K3](/models/kimi-k3) | Moonshot AI | 72% |
| 23 | [GLM-5.3](/models/glm-5-3) | Z.AI | 72% |
| 24 | [Qwen3.7 Max](/models/qwen3-7-max) | Alibaba | 72% |
| 25 | [Qwen3.6 Plus](/models/qwen3-6-plus) | Alibaba | 72% |
| 26 | [Claude Haiku 4.5](/models/claude-haiku-4-5) | Anthropic | 71% |
| 27 | [Claude Opus 5](/models/claude-opus-5) | Anthropic | 70% |
| 28 | [GPT-6 Astra](/models/gpt-6-astra) | OpenAI | 69% |
| 29 | [Qwen3.5 397B](/models/qwen3-5-397b) | Alibaba | 69% |
| 30 | [Grok 4.20 Multi-agent](/models/grok-4-20-multi-agent-beta) | xAI | 67% |
| 31 | [GLM-5.3](/models/glm-5-3) | Z.AI | 66% |
| 32 | [Gemini 3.5 Flash-Lite](/models/gemini-3-5-flash-lite) | Google | 66% |
| 33 | [Grok 4.6](/models/grok-4-6) | xAI | 66% |
| 34 | [Kimi K2.6](/models/kimi-2-6) | Moonshot AI | 65% |
| 35 | [stealth/ox-alpha (low)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Stealth | 65% |
| 36 | [Claude Fable 5.1](/models/claude-fable-5-1) | Anthropic | 64% |
| 37 | [GPT-6 Astra](/models/gpt-6-astra) | OpenAI | 64% |
| 38 | [MiniMax M3](/models/minimax-m3) | MiniMax | 64% |
| 39 | [Grok 4.20 Multi-agent](/models/grok-4-20-multi-agent-beta) | xAI | 64% |
| 40 | [Qwen3.6 Plus](/models/qwen3-6-plus) | Alibaba | 63% |
| 41 | [MiniMax M3](/models/minimax-m3) | MiniMax | 63% |
| 42 | [MiMo-V2.5-Pro](/models/mimo-v2-5-pro) | Xiaomi | 62% |
| 43 | [Gemini 3.5 Flash-Lite](/models/gemini-3-5-flash-lite) | Google | 60% |
| 44 | [Qwen3.6 Plus](/models/qwen3-6-plus) | Alibaba | 59% |
| 45 | [DeepSeek V4.1 Flash](/models/deepseek-v4-1-flash) | DeepSeek | 57% |
| 46 | [DeepSeek V4.1 Flash](/models/deepseek-v4-1-flash) | DeepSeek | 57% |
| 47 | [Qwen3.7 Max](/models/qwen3-7-max) | Alibaba | 57% |
| 48 | [Claude Fable 5](/models/claude-fable) | Anthropic | 56% |
| 49 | [Grok 4.6](/models/grok-4-6) | xAI | 56% |
| 50 | [Grok 4.20](/models/grok-4-20-beta) | xAI | 56% |
| 51 | [Grok 4.5](/models/grok-4-5) | xAI | 55% |
| 52 | [stealth/ox-alpha (max)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Stealth | 54% |
| 53 | [GPT-5.6 Terra](/models/gpt-5-6-terra) | OpenAI | 54% |
| 54 | [Grok 4.5](/models/grok-4-5) | xAI | 54% |
| 55 | [Grok 4.20](/models/grok-4-20-beta) | xAI | 54% |
| 56 | [Nemotron 3 Super 120B A12B](/models/nemotron-3-super-120b-a12b) | NVIDIA | 54% |
| 57 | [Kimi K2.5](/models/kimi-k2-5) | Moonshot AI | 52% |
| 58 | [Grok 4.3](/models/grok-4-3) | xAI | 50% |
| 59 | [Kimi K2.6](/models/kimi-2-6) | Moonshot AI | 50% |
| 60 | [Muse Spark 1.2](/models/muse-spark-1-2) | Meta | 50% |
| 61 | [anthropic/claude-3.5-haiku](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Anthropic | 50% |
| 62 | [Muse Spark 1.2](/models/muse-spark-1-2) | Meta | 49% |
| 63 | [anthropic/claude-3.7-sonnet:thinking](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Anthropic | 49% |
| 64 | [nvidia/nemotron-3-ultra-550b-a55b (none)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | NVIDIA | 49% |
| 65 | [GPT-5.6 Sol](/models/gpt-5-6-sol) | OpenAI | 48% |
| 66 | [GPT-5.4](/models/gpt-5-4) | OpenAI | 48% |
| 67 | [Gemini 3 Pro](/models/gemini-3-pro) | Google | 48% |
| 68 | [GPT-5.6 Sol](/models/gpt-5-6-sol) | OpenAI | 47% |
| 69 | [GPT-5.5](/models/gpt-5-5) | OpenAI | 47% |
| 70 | [GPT-5.6 Terra](/models/gpt-5-6-terra) | OpenAI | 47% |
| 71 | [Nemotron 3 Super 120B A12B](/models/nemotron-3-super-120b-a12b) | NVIDIA | 47% |
| 72 | [Qwen3.6 Plus](/models/qwen3-6-plus) | Alibaba | 46% |
| 73 | [Grok 4.3](/models/grok-4-3) | xAI | 46% |
| 74 | [Claude 3.5 Sonnet](/models/claude-3-5-sonnet) | Anthropic | 46% |
| 75 | [Claude Fable 5](/models/claude-fable) | Anthropic | 45% |
| 76 | [GPT-5.5](/models/gpt-5-5) | OpenAI | 45% |
| 77 | [GPT-5.5](/models/gpt-5-5) | OpenAI | 45% |
| 78 | [GPT-5.2-Codex](/models/gpt-5-2-codex) | OpenAI | 45% |
| 79 | [GPT-5.1](/models/gpt-5-1) | OpenAI | 45% |
| 80 | [anthropic/claude-3.7-sonnet](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Anthropic | 44% |
| 81 | [Claude 4.1 Opus](/models/claude-4-1-opus) | Anthropic | 43% |
| 82 | [Nemotron 3 Super 120B A12B](/models/nemotron-3-super-120b-a12b) | NVIDIA | 43% |
| 83 | [openrouter/hunter-alpha (none)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Stealth | 43% |
| 84 | [GPT-5.4](/models/gpt-5-4) | OpenAI | 42% |
| 85 | [Claude 4.1 Opus](/models/claude-4-1-opus) | Anthropic | 42% |
| 86 | [GPT-5.6 Luna](/models/gpt-5-6-luna) | OpenAI | 41% |
| 87 | [GPT-5.3 Instant](/models/gpt-5-3-instant) | OpenAI | 40% |
| 88 | [nvidia/nemotron-3-ultra-550b-a55b (xhigh)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | NVIDIA | 40% |
| 89 | [DeepSeek V4 Flash 0731](/models/deepseek-v4-flash-0731) | DeepSeek | 39% |
| 90 | [GPT-5 Codex](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | — | 39% |
| 91 | [GPT-5.2-Codex](/models/gpt-5-2-codex) | OpenAI | 39% |
| 92 | [Gemini 3.6 Flash](/models/gemini-3-6-flash) | Google | 39% |
| 93 | [GPT-5.2](/models/gpt-5-2) | OpenAI | 38% |
| 94 | [MiMo-V2.5-Pro](/models/mimo-v2-5-pro) | Xiaomi | 38% |
| 95 | [GPT-5.6 Luna](/models/gpt-5-6-luna) | OpenAI | 37% |
| 96 | [Gemini 3.1 Pro](/models/gemini-3-1-pro) | Google | 37% |
| 97 | [GPT-5.2-Codex](/models/gpt-5-2-codex) | OpenAI | 37% |
| 98 | [openrouter/healer-alpha (none)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Stealth | 37% |
| 99 | [GPT-5.5 Pro](/models/gpt-5-5-pro) | OpenAI | 36% |
| 100 | [Gemini 3 Pro Deep Think](/models/gemini-3-pro-deep-think) | Google | 36% |
| 101 | [DeepSeek V4 Pro 0813](/models/deepseek-v4-pro-0813) | DeepSeek | 35% |
| 102 | [Gemini 3.7 Flash](/models/gemini-3-7-flash) | Google | 35% |
| 103 | [MiMo-V2.5](/models/mimo-v2-5) | Xiaomi | 35% |
| 104 | [openrouter/hunter-alpha (xhigh)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Stealth | 35% |
| 105 | [GPT-5.5 Pro](/models/gpt-5-5-pro) | OpenAI | 34% |
| 106 | [GPT-5.5](/models/gpt-5-5) | OpenAI | 34% |
| 107 | [Claude Opus 4](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | — | 34% |
| 108 | [GPT-5.4 mini](/models/gpt-5-4-mini) | OpenAI | 32% |
| 109 | [GPT-5.4 mini](/models/gpt-5-4-mini) | OpenAI | 32% |
| 110 | [DeepSeek V4 Pro 0813](/models/deepseek-v4-pro-0813) | DeepSeek | 32% |
| 111 | [GPT-5.1-Codex-Max](/models/gpt-5-1-codex-max) | OpenAI | 32% |
| 112 | [GPT-5.4 mini](/models/gpt-5-4-mini) | OpenAI | 31% |
| 113 | [Kimi K2.5 (Reasoning)](/models/kimi-k2-5-reasoning) | Moonshot AI | 31% |
| 114 | [Gemini 3.1 Pro](/models/gemini-3-1-pro) | Google | 31% |
| 115 | [Gemini 3.7 Flash](/models/gemini-3-7-flash) | Google | 31% |
| 116 | [GLM-5-Turbo](/models/glm-5-turbo) | Z.AI | 31% |
| 117 | [Nemotron 3 Super 120B A12B](/models/nemotron-3-super-120b-a12b) | NVIDIA | 31% |
| 118 | [GLM-5.2](/models/glm-5-2) | Z.AI | 31% |
| 119 | [Claude 4 Sonnet](/models/claude-4-sonnet) | Anthropic | 30% |
| 120 | [Qwen3.8-27B](/models/qwen3-8-27b) | Alibaba | 29% |
| 121 | [Claude 4 Sonnet](/models/claude-4-sonnet) | Anthropic | 29% |
| 122 | [Llama 4 Maverick](/models/llama-4-maverick) | Meta | 29% |
| 123 | [GPT-5.2](/models/gpt-5-2) | OpenAI | 28% |
| 124 | [Gemini 3.6 Flash](/models/gemini-3-6-flash) | Google | 28% |
| 125 | [GLM-5 (Reasoning)](/models/glm-5-reasoning) | Z.AI | 28% |
| 126 | [Nemotron 3 Nano 30B A3B (none)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | — | 28% |
| 127 | [Qwen3.8 Max](/models/qwen3-8-max) | Alibaba | 27% |
| 128 | [GPT-5.2 Instant](/models/gpt-5-2-instant) | OpenAI | 27% |
| 129 | [o3](/models/o3) | OpenAI | 26% |
| 130 | [openrouter/healer-alpha (xhigh)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Stealth | 26% |
| 131 | [GPT-5.1](/models/gpt-5-1) | OpenAI | 25% |
| 132 | [Gemma 4 31B](/models/gemma-4-31b) | Google | 25% |
| 133 | [GPT-5.3 Codex](/models/gpt-5-3-codex) | OpenAI | 24% |
| 134 | [MiMo-V2.5](/models/mimo-v2-5) | Xiaomi | 24% |
| 135 | [GLM-5-Turbo](/models/glm-5-turbo) | Z.AI | 23% |
| 136 | [GLM-5.1](/models/glm-5-1) | Z.AI | 22% |
| 137 | [Step 3.5 Flash](/models/step-3-5-flash) | StepFun | 22% |
| 138 | [Laguna S 2.1](/models/laguna-s-2-1) | Poolside | 22% |
| 139 | [GPT-5](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | — | 21% |
| 140 | [Gemma 4 26B A4B](/models/gemma-4-26b-a4b) | Google | 21% |
| 141 | [Laguna S 2.1](/models/laguna-s-2-1) | Poolside | 21% |
| 142 | [GPT-5.3 Codex](/models/gpt-5-3-codex) | OpenAI | 20% |
| 143 | [DeepSeek V4 Flash 0731](/models/deepseek-v4-flash-0731) | DeepSeek | 20% |
| 144 | [Qwen3 Coder 480B A35B](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | — | 20% |
| 145 | [Gemini 2.5 Pro](/models/gemini-2-5-pro) | Google | 20% |
| 146 | [GLM-5](/models/glm-5) | Z.AI | 20% |
| 147 | [Llama 4 Scout](/models/llama-4-scout) | Meta | 20% |
| 148 | [Gemma 4 31B](/models/gemma-4-31b) | Google | 20% |
| 149 | [Gemini 3.5 Flash](/models/gemini-3-5-flash) | Google | 20% |
| 150 | [GPT-5.3 Codex](/models/gpt-5-3-codex) | OpenAI | 19% |
| 151 | [Grok 4.1 Fast](/models/grok-4-1-fast) | xAI | 19% |
| 152 | [Gemini 2.5 Flash](/models/gemini-2-5-flash) | Google | 19% |
| 153 | [Gemini 3.5 Flash](/models/gemini-3-5-flash) | Google | 19% |
| 154 | [GPT-5](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | — | 18% |
| 155 | [DeepSeek V4 Flash 0731](/models/deepseek-v4-flash-0731) | DeepSeek | 18% |
| 156 | [GLM-5.1](/models/glm-5-1) | Z.AI | 18% |
| 157 | [GLM-5.2](/models/glm-5-2) | Z.AI | 17% |
| 158 | [Trinity-Large-Thinking](/models/trinity-large-thinking) | Arcee AI | 17% |
| 159 | [MiMo-V2-Flash](/models/mimo-v2-flash) | Xiaomi | 16% |
| 160 | [Hy3](/models/hy3) | Tencent | 16% |
| 161 | [google/gemini-2.0-flash-001](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Google | 15% |
| 162 | [DeepSeek V4 Pro 0813](/models/deepseek-v4-pro-0813) | DeepSeek | 14% |
| 163 | [meta-llama/llama-3.1-8b-instruct](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Meta | 14% |
| 164 | [GPT-5.4 nano](/models/gpt-5-4-nano) | OpenAI | 14% |
| 165 | [GPT-5.4 nano](/models/gpt-5-4-nano) | OpenAI | 14% |
| 166 | [DeepSeek V4 Pro 0813](/models/deepseek-v4-pro-0813) | DeepSeek | 14% |
| 167 | [GPT-4.1](/models/gpt-4-1) | OpenAI | 14% |
| 168 | [DeepSeek V4 Flash 0731](/models/deepseek-v4-flash-0731) | DeepSeek | 14% |
| 169 | [DeepSeek V3.2 (Thinking)](/models/deepseek-v3-2-thinking) | DeepSeek | 13% |
| 170 | [Step 3.5 Flash](/models/step-3-5-flash) | StepFun | 13% |
| 171 | [Trinity-Large-Thinking](/models/trinity-large-thinking) | Arcee AI | 13% |
| 172 | [MiMo-V2-Flash](/models/mimo-v2-flash) | Xiaomi | 13% |
| 173 | [GPT-4o](/models/gpt-4o) | OpenAI | 12% |
| 174 | [GPT-OSS 120B](/models/gpt-oss-120b) | OpenAI | 12% |
| 175 | [Gemma 4 26B A4B](/models/gemma-4-26b-a4b) | Google | 11% |
| 176 | [Gemini 3.1 Flash-Lite](/models/gemini-3-1-flash-lite) | Google | 11% |
| 177 | [Seed 1.6](/models/seed-1-6) | ByteDance | 11% |
| 178 | [baidu/ernie-4.5-vl-424b-a47b (xhigh)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Baidu | 11% |
| 179 | [GPT-5.4 nano](/models/gpt-5-4-nano) | OpenAI | 10% |
| 180 | [Gemini 3 Flash](/models/gemini-3-flash) | Google | 10% |
| 181 | [DeepSeek V3.2](/models/deepseek-v3-2) | DeepSeek | 10% |
| 182 | [Claude 3 Haiku](/models/claude-3-haiku) | Anthropic | 10% |
| 183 | [Gemini 3 Flash](/models/gemini-3-flash) | Google | 10% |
| 184 | [nvidia/nemotron-3-nano-30b-a3b:free (xhigh)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | NVIDIA | 10% |
| 185 | [Kimi K2](/models/kimi-k2) | Moonshot AI | 10% |
| 186 | [Grok 4.1 Fast](/models/grok-4-1-fast) | xAI | 10% |
| 187 | [MiniMax M2.5](/models/minimax-m2-5) | MiniMax | 9% |
| 188 | [Hy3](/models/hy3) | Tencent | 8% |
| 189 | [MiniMax M2.5](/models/minimax-m2-5) | MiniMax | 8% |
| 190 | [GLM-4.5](/models/glm-4-5) | Z.AI | 8% |
| 191 | [MiniMax M2.7](/models/minimax-m2-7) | MiniMax | 8% |
| 192 | [DeepSeek-R1](/models/deepseek-r1) | DeepSeek | 8% |
| 193 | [o4-mini (high)](/models/o4-mini-high) | OpenAI | 8% |
| 194 | [Seed 1.6](/models/seed-1-6) | ByteDance | 7% |
| 195 | [MiniMax M2.7](/models/minimax-m2-7) | MiniMax | 7% |
| 196 | [DeepSeek-R1](/models/deepseek-r1) | DeepSeek | 7% |
| 197 | [prime-intellect/intellect-3 (low)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Prime Intellect | 7% |
| 198 | [mistralai/mistral-small-2603 (high)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Mistral | 6% |
| 199 | [qwen/qwen3-235b-a22b (none)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Alibaba | 6% |
| 200 | [GLM-4.5](/models/glm-4-5) | Z.AI | 6% |
| 201 | [GPT-OSS 120B](/models/gpt-oss-120b) | OpenAI | 5% |
| 202 | [nvidia/nemotron-nano-9b-v2:free (none)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | NVIDIA | 5% |
| 203 | [prime-intellect/intellect-3 (high)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Prime Intellect | 5% |
| 204 | [ai21/jamba-large-1.7](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | AI21 Labs | 5% |
| 205 | [o4-mini (high)](/models/o4-mini-high) | OpenAI | 4% |
| 206 | [baidu/ernie-4.5-300b-a47b](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Baidu | 4% |
| 207 | [deepseek/deepseek-chat](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | DeepSeek | 4% |
| 208 | [mistralai/mistral-small-2603 (none)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Mistral | 4% |
| 209 | [baidu/ernie-4.5-vl-424b-a47b (none)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Baidu | 3% |
| 210 | [qwen/qwen3-235b-a22b (xhigh)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Alibaba | 3% |
| 211 | [nvidia/nemotron-nano-9b-v2:free (xhigh)](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | NVIDIA | 3% |
| 212 | [google/gemma-3-27b-it](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Google | 3% |
| 213 | [mistralai/mistral-large-2512](https://petergpt.github.io/bullshit-benchmark/viewer/index.v2.html) | Mistral | 2% |
| 214 | [GPT-4o mini](/models/gpt-4o-mini) | OpenAI | 2% |

## FAQ

### What does BullshitBench v2 measure?

A benchmark that tests whether AI models challenge nonsensical, ill-posed, or logically flawed prompts instead of confidently generating incorrect answers. Measures the critical ability to push back on bad input.

### Which model leads the published BullshitBench v2 snapshot?

Claude Opus 4.8 currently leads the published BullshitBench v2 snapshot with a score of 95%.

### How many models are evaluated on BullshitBench v2?

The September 10, 2026 at 8:20 PM UTC contains 214 AI models.

### Does BullshitBench v2 affect BenchLM's overall score?

Not directly. BullshitBench v2 is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.