VECTOR WIREAI INTELLIGENCE
NVDA$1,847+3.2%MSFT$512+1.1%GOOGL$199-0.4%META$728+2.7%AMD$184-1.2%TSM$212+0.6%PLTR$98+4.1%AI IDX4,821+1.9%
PKT
SEEDRefresh Models Deals Regulatory Sources

arc_challenge

48 models tracked

AI2 ARC-Challenge — Classic AI2 Reasoning Challenge (science QA) — unrelated to ARC-AGI despite shared 'ARC' acronym

#ModelVendorBest scoreRunsLast seen
1Nemotron 3 Ultra 550B A55B BaseNVIDIA97.312026-06-05
2NVIDIA-Nemotron-3-Ultra-550B-A55B-Base-BF16NVIDIA97.312026-08-24
3Mistral Large 3 675B Base 2512Mistral97.312026-06-05
4MiMo V2.5 Pro BaseXiaomi97.232026-06-05
5MiMo V2.5 ProXiaomi97.212026-08-23
6Llama 3.1 Instruct 405BMeta96.942026-08-23
7Kimi K2 BaseMoonshot96.782026-06-13
8GLM-4.5-BaseZ.ai96.632026-06-05
9MiMo V2.5 BaseXiaomi96.532026-06-05
10Claude 3 OpusAnthropic96.412026-08-23
11Nemotron 3 Super 120B A12BNVIDIA96.112026-08-11
12Hy3 preview-BaseTencent9622026-06-05
13MiMo V2 Flash BaseXiaomi95.922026-06-13
14DeepSeek-V3.1-BaseDeepSeek95.622026-06-13
15DeepSeek-V3.2-Exp-BaseDeepSeek95.532026-06-13
16Qwen3.5 35B A3BAlibaba95.412026-08-11
17DeepSeek-V3DeepSeek95.312026-05-01
18Hunyuan LargeTencent9512026-05-31
19Llama 3.1 70B InstructMeta94.822026-08-23
20Nova Pro (Non-Reasoning)Amazon94.812026-08-23
21DeepSeek-V3-BaseDeepSeek94.722026-06-05
22Hunyuan Large Inst.Tencent94.612026-05-31
23Qwen2.5 Instruct 72BAlibaba94.512026-05-01
24Claude 3 SonnetAnthropic93.212026-08-23
25Jamba 1.5 LargeAI219312026-08-23
26Gemma 4 26B A4BGoogle92.812026-08-11
27NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16NVIDIA92.712026-08-24
28Nemotron 3.5 LightningNVIDIA92.712026-08-11
29Nova Lite (Non-Reasoning)Amazon92.412026-08-23
30DeepSeek-V2DeepSeek92.422026-05-31
31NVIDIA Nemotron 3 Nano 30B A3BNVIDIA9212026-08-11
32Phi 3.5 MoE InstructMicrosoft9112026-08-23
33Claude 2Anthropic9112026-07-17
34Nova Micro (Non-Reasoning)Amazon90.212026-08-23
35Claude 1.3Anthropic9012026-07-17
36Claude 3 HaikuAnthropic89.212026-08-23
37Jamba 1.5 MiniAI2185.712026-08-23
38Phi 3.5 Mini InstructMicrosoft84.612026-08-23
39Phi 4 Mini InstructMicrosoft83.712026-08-23
40Llama 3.1 8B InstructMeta83.412026-08-23
41Llama 3.2 3B InstructMeta78.612026-08-23
42Gemma 2 27B ITGoogle71.412026-08-23
43Command R+ (Apr '24)Cohere7112026-08-23
44Qwen 2.5 Coder 32B InstructAlibaba70.512026-08-23
45Llama 3.1 Nemotron Instruct 70BNVIDIA69.212026-08-23
46Qwen2 Instruct (72B)Alibaba68.912026-08-23
47Gemma 2 Instruct (9B)Google68.412026-08-23
48ERNIE 4.5Baidu40.612026-08-23

Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.