Skip to main content

Best AI models for tape-machine simulation

Ranked by accuracy on the meo “tape-machine simulation” domain — one of eleven un-leaked, contamination-resistant domains. Per-domain strength reveals where a model genuinely reasons versus where it pattern-matches; the leaders differ sharply by domain.

Ranking 82 models across the meo-tested roster · as of 2026-10-18.

#ModelLabaccuracy
1OpenAI: GPT-5.5openai100%
2Google: Gemini 4 Argongoogle100%
3OpenAI: GPT-5.5 Proopenai100%
4OpenAI: GPT-6.1 Solopenai100%
5Anthropic: Claude Opus 4.8anthropic100%
6Google: Gemini 3.5 Flashgoogle100%
7DeepSeek: DeepSeek V4 Flash 0423deepseek100%
8Qwen: Qwen3.7 Maxqwen100%
9MoonshotAI: Kimi K3moonshotai100%
10Google: Gemini 3.1 Pro Previewgoogle100%
11NVIDIA: Nemotron 3 Ultranvidia100%
12SpaceXAI: Grok 4.7x-ai100%
13Xiaomi: MiMo-V2.5-Proxiaomi100%
14OpenAI: GPT-5.4 Nanoopenai100%
15Qwen: Qwen3.6 Flashqwen100%
16Qwen: Qwen3.5-Flashqwen100%
17MoonshotAI: Kimi K2.6moonshotai100%
18OpenAI: GPT-5 Miniopenai100%
19OpenAI: GPT-5.1-Codex-Miniopenai100%
20SpaceXAI: Grok 4.3x-ai100%
21Qwen: Qwen3 VL 32B Instructqwen100%
22ByteDance Seed: Seed 1.6bytedance-seed100%
23Qwen: Qwen3 VL 235B A22B Instructqwen100%
24Google: Gemma 4 31Bgoogle100%
25OpenAI: GPT-4.1 Miniopenai100%
26Qwen: Qwen3.5 Plus 2026-02-15qwen100%
27Mistral: Mistral Large 3 2512mistralai100%
28Perceptron: Perceptron Mk1perceptron100%
29Anthropic: Claude Opus 5.5anthropic93%
30Anthropic: Claude Fable 5anthropic93%
31inclusionAI: Ring-2.6-1Tinclusionai93%
32Z.ai: GLM 5.2z-ai93%
33DeepSeek: DeepSeek V4 Pro 0423deepseek93%
34MiniMax: MiniMax M3minimax93%
35MoonshotAI: Kimi K2.7 Codemoonshotai93%
36Qwen: Qwen3.7 Plusqwen93%
37Nex AGI: Nex-N2-Pro (free)nex-agi93%
38Qwen: Qwen3 VL 30B A3B Thinkingqwen93%
39Google: Gemini 3.1 Flash Lite Previewgoogle93%
40Mistral: Ministral 3 8B 2512mistralai93%
41Mistral: Mistral Small 3.2 24Bmistralai93%
42Qwen: Qwen3.5-27Bqwen93%
43Owl Alphaopenrouter87%
44Xiaomi: MiMo-V2.5xiaomi87%
45Qwen: Qwen3 VL 8B Thinkingqwen87%
46Google: Gemini 2.5 Flash Litegoogle87%
47Mistral: Ministral 3 14B 2512mistralai87%
48Google: Gemma 3 27Bgoogle87%
49Meta: Llama 4 Scoutmeta-llama87%
50Baidu: ERNIE 4.5 VL 424B A47B baidu80%
51Google: Gemma 4 26B A4B google80%
52Meta: Llama 4 Maverickmeta-llama80%
53MiniMax: MiniMax-01minimax80%
54Qwen: Qwen3 VL 30B A3B Instructqwen73%
55Qwen: Qwen2.5 VL 72B Instructqwen73%
56Z.ai: GLM 4.6Vz-ai67%
57ByteDance Seed: Seed 1.6 Flashbytedance-seed67%
58Qwen: Qwen3 VL 8B Instructqwen67%
59Google: Gemma 3 12Bgoogle67%
60OpenAI: GPT-4o-miniopenai67%
61Anthropic: Claude 3 Haikuanthropic67%
62Z.ai: GLM 5.1z-ai53%
63Mistral: Mistral Small 4mistralai53%
64OpenAI: GPT-5 Nanoopenai53%
65OpenAI: GPT-4.1 Nanoopenai53%
66Mistral: Ministral 3 3B 2512mistralai53%
67Google: Gemma 3 4Bgoogle40%
68Meta: Llama 3.2 11B Vision Instructmeta-llama40%
69Tencent: Hy3 previewtencent33%
70Perplexity: Sonarperplexity33%
71Amazon: Nova Lite 1.0amazon33%
72Google: Gemini 3.1 Flash Litegoogle27%
73ByteDance Seed: Seed-2.0-Litebytedance-seed27%
74NVIDIA: Nemotron 3 Nano Omni (free)nvidia7%
75Qwen: Qwen3.6 Plusqwen7%
76Qwen: Qwen3.5-35B-A3Bqwen7%
77StepFun: Step 3.7 Flashstepfun0%
78Qwen: Qwen3.6 35B A3Bqwen0%
79NVIDIA: Nemotron Nano 12B 2 VL (free)nvidia0%
80Arcee AI: Trinity Large Thinkingarcee-ai0%
81Qwen: Qwen3.5-9Bqwen0%
82ByteDance Seed: Seed-2.0-Minibytedance-seed0%

← All rankingsMethodology & 𝕍 →