AgMoDB
ModelsAgentsOrchestratorsEvalsCompositesVisualizeIndustry
AgMoDB by @mistakeknot
All Benchmarks

ARC-AGI-2

ARC-AGI-2 is the harder successor to ARC-AGI-1 for abstract puzzle generalization. It is aimed at separating frontier systems after v1 saturation trends.

reasoningarc_agi

Strengths

Harder task distribution preserves separation among frontier systems; semi-private setup continues to limit straightforward benchmark gaming.

Caveats

Still measures a narrow abstraction domain rather than end-to-end product utility; reported outcomes can shift with retry budgets and evaluation harness choices.

How to interpret scores

Contamination risk: lowFreshness: static

Higher is better. Percentage of tasks solved correctly.

Max score: 100

Relevant use cases

Abstract reasoningAGI researchCognitive evaluationPattern RecognitionNovel Problem Solving

Caveat to keep in mind

Still measures a narrow abstraction domain rather than end-to-end product utility.

Leaderboard

Last synced: Nov 22, 2026, 12:17 PM
#ModelProviderScorePrice/M
1
Gemma 3 270M
source benchpress
Evidence observed Apr 13, 2026
Google100.0$0.00
2
Llama 2 Chat 7B
source benchpress
Evidence observed Apr 13, 2026
Meta100.0$0.00
3
Qwen3.5 0.8B (Reasoning)
source benchpress
Evidence observed Apr 13, 2026
Alibaba99.9$0.00
4
Mistral: Ministral 3 3B 2512
source benchpress
Evidence observed Apr 13, 2026
Mistral AI99.7$0.10
5
AlfredPros: CodeLLaMa 7B Instruct Solidity
source benchpress
Evidence observed Apr 13, 2026
Alfredpros99.5$0.90
6
Mistral: Ministral 3 8B 2512
source benchpress
Evidence observed Apr 13, 2026
Mistral AI98.8$0.15
7
Qwen3.5 9B (Reasoning)
source benchpress
Evidence observed Apr 13, 2026
Alibaba97.1$0.15
8
Mistral: Ministral 3 14B 2512
source benchpress
Evidence observed Apr 13, 2026
Mistral AI95.8$0.20
9
GPT-5.2 Codex (Xhigh)
source benchpress
Evidence observed Apr 13, 2026
OpenAI94.6$4.81
10
Gemini 4 Carbon (Medium)
source arc_agi
Evidence observed Nov 22, 2026
Google94.2$8.00
11
GPT-6.1 Sol (Max)
source arc_agi
Evidence observed Nov 22, 2026
OpenAI94.2$4.00
12
GPT-6 Astra (Xhigh)
source arc_agi
Evidence observed Nov 22, 2026
OpenAI93.3$20.00
13
Claude Opus 5.5 (High, Default Fallback)
source arc_agi
Evidence observed Oct 2, 2026
Anthropic93.3$8.00
14
KAT-Coder-Pro V1
source benchpress
Evidence observed Apr 13, 2026
KwaiKAT92.9$0.00
15
Qwen3 Max Thinking
source benchpress
Evidence observed Apr 13, 2026
Alibaba92.7$0.00
16
Claude Opus 5.5 (Xhigh, Default Fallback)
source arc_agi
Evidence observed Oct 2, 2026
Anthropic92.5$8.00
17
GPT-5.6 Sol (Max)
source arc_agi
Evidence observed Nov 22, 2026
OpenAI92.5$8.00
18
GPT-6 Astra (High)
source arc_agi
Evidence observed Nov 22, 2026
OpenAI92.1$20.00
19
GPT-6 Astra (Medium)
source arc_agi
Evidence observed Nov 22, 2026
OpenAI92.1$20.00
20
Gemini 4 Carbon (High)
source arc_agi
Evidence observed Nov 22, 2026
Google92.1$8.00
21
GPT-6.1 Sol (Xhigh)
source arc_agi
Evidence observed Nov 22, 2026
OpenAI91.7$4.00
22
OpenAI: GPT-6.2 Astra
source arc_agi
Evidence observed Nov 22, 2026
OpenAI91.7$16.00
23
GPT-6.1 Sol (High)
source arc_agi
Evidence observed Nov 22, 2026
OpenAI91.7$4.00
24
Claude Opus 5 (Max)
source arc_agi
Evidence observed Nov 22, 2026
Anthropic90.4$10.00
25
GPT-5.6 Sol (Xhigh)
source arc_agi
Evidence observed Nov 22, 2026
OpenAI90.0$8.00
26
Qwen3.5 0.8B (Non-reasoning)
source benchpress
Evidence observed Apr 13, 2026
Alibaba89.4$0.00
27
Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 5 Fallback)
source arc_agi
Evidence observed Nov 22, 2026
Anthropic89.2$20.00
28
Gemini 3.8 Flash (High)
source arc_agi
Evidence observed Nov 22, 2026
Google89.2$1.50
29
Claude Opus 5 (High)
source arc_agi
Evidence observed Nov 22, 2026
Anthropic88.3$10.00
30
Claude Opus 5.5 (Medium, Default Fallback)
source arc_agi
Evidence observed Oct 2, 2026
Anthropic87.5$8.00
31
GPT-6.1 Sol (Medium)
source arc_agi
Evidence observed Nov 22, 2026
OpenAI86.7$4.00
32
Anthropic: Claude Fable 5.2
source arc_agi
Evidence observed Nov 22, 2026
Anthropic86.7$16.00
33
GPT-5.6 Sol (High)
source arc_agi
Evidence observed Nov 22, 2026
OpenAI85.4$8.00
34
GPT-6 Astra (Low)
source arc_agi
Evidence observed Nov 22, 2026
OpenAI85.4$20.00
35
GPT-5.5 (Xhigh)
source arc_agi
Evidence observed Nov 22, 2026
OpenAI85.0$11.25
36
Gemini 3.7 Flash (High)
source arc_agi
Evidence observed Nov 22, 2026
Google84.6$1.50
37
Gemini 3 Deep Think
source arc_agi
Evidence observed Nov 22, 2026
Google84.6$0.00
38
Ministral 3 3B
source benchpress
Evidence observed Apr 13, 2026
Mistral AI84.4$0.10
39
GPT-5.5 Pro (Xhigh)
source arc_agi
Evidence observed Nov 22, 2026
OpenAI84.2$33.75
40
GPT-5.4 Pro (Xhigh)
source arc_agi
Evidence observed Nov 22, 2026
OpenAI83.3$67.50
41
GPT-5.5 (High)
source arc_agi
Evidence observed Nov 22, 2026
OpenAI83.3$11.25
42
Gemini 3.8 Flash (Medium)
source arc_agi
Evidence observed Nov 22, 2026
Google82.9$1.50
43
GPT-5.6 Terra (Max)
source arc_agi
Evidence observed Nov 22, 2026
OpenAI82.5$4.50
44
GPT-5.3 Codex (Xhigh)
source benchpress
Evidence observed Apr 13, 2026
OpenAI82.1$4.81
45
Qwen3.5 9B (Non-reasoning)
source benchpress
Evidence observed Apr 13, 2026
Alibaba79.3$0.19
46
Gemma 3 1B Instruct
source benchpress
Evidence observed Apr 13, 2026
Google79.2$0.00
47
Anthropic: Claude Fable 5.1
source arc_agi
Evidence observed Nov 22, 2026
Anthropic78.3$20.00
48
GPT-6 Sol (Xhigh)
source arc_agi
Evidence observed Nov 22, 2026
OpenAI78.1$4.00
49
LFM2.5-1.2B-Instruct
source benchpress
Evidence observed Apr 13, 2026
Liquid AI78.0$0.00
50
Gemini 3.8 Flash (Low)
source arc_agi
Evidence observed Nov 22, 2026
Google77.5$1.50