{"benchmark_id":"arc-agi","benchmark_name":"ARC-AGI","benchmark_description":"The Abstraction and Reasoning Corpus for Artificial General Intelligence (ARC-AGI) is a benchmark designed to test general intelligence and abstract reasoning capabilities through visual grid-based transformation tasks. Each task consists of 2-5 demonstration pairs showing input grids transformed into output grids according to underlying rules, with test-takers required to infer these rules and apply them to novel test inputs. The benchmark uses colored grids (up to 30x30) with 10 discrete colors/symbols, designed to measure human-like general fluid intelligence and skill-acquisition efficiency with minimal prior knowledge.","max_score":1.0,"categories":["reasoning","spatial_reasoning","vision"],"modality":"image","total_models":8,"entries":[{"rank":1,"model_id":"gpt-5.5","model_name":"GPT-5.5","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.95,"normalized_score":0.95,"verified":false,"self_reported":true,"provider_id":"openai","input_cost_per_million":5.0,"output_cost_per_million":30.0,"speed_rps":null,"context_window":1050000,"release_date":"2026-04-23","announcement_date":"2026-04-23","multimodal":true,"param_count":null,"is_new":false},{"rank":2,"model_id":"gpt-5.4","model_name":"GPT-5.4","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.937,"normalized_score":0.937,"verified":false,"self_reported":true,"provider_id":"openai","input_cost_per_million":2.5,"output_cost_per_million":15.0,"speed_rps":50.0,"context_window":1000000,"release_date":"2026-03-05","announcement_date":"2026-03-05","multimodal":true,"param_count":null,"is_new":false},{"rank":3,"model_id":"gpt-5.2-pro-2025-12-11","model_name":"GPT-5.2 Pro","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.905,"normalized_score":0.905,"verified":false,"self_reported":true,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"speed_rps":null,"context_window":null,"release_date":"2025-12-11","announcement_date":"2025-12-11","multimodal":true,"param_count":null,"is_new":false},{"rank":4,"model_id":"o3-2025-04-16","model_name":"o3","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.88,"normalized_score":0.88,"verified":false,"self_reported":true,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"speed_rps":null,"context_window":null,"release_date":"2025-04-16","announcement_date":"2025-04-16","multimodal":true,"param_count":null,"is_new":false},{"rank":5,"model_id":"gpt-5.2-2025-12-11","model_name":"GPT-5.2","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.862,"normalized_score":0.862,"verified":false,"self_reported":true,"provider_id":"openai","input_cost_per_million":1.75,"output_cost_per_million":14.0,"speed_rps":100.0,"context_window":400000,"release_date":"2025-12-11","announcement_date":"2025-12-11","multimodal":true,"param_count":null,"is_new":false},{"rank":6,"model_id":"inkling-small","model_name":"Inkling-Small","organization_name":"Thinking Machines Lab","organization_id":"thinking-machines","benchmark_score":0.84,"normalized_score":0.84,"verified":false,"self_reported":true,"provider_id":"thinking-machines","input_cost_per_million":0.3,"output_cost_per_million":1.2,"speed_rps":null,"context_window":256000,"release_date":"2026-07-30","announcement_date":"2026-07-30","multimodal":true,"param_count":276000000000,"is_new":false},{"rank":7,"model_id":"longcat-flash-thinking","model_name":"LongCat-Flash-Thinking","organization_name":"Meituan","organization_id":"meituan","benchmark_score":0.503,"normalized_score":0.503,"verified":false,"self_reported":true,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"speed_rps":null,"context_window":null,"release_date":"2025-09-22","announcement_date":"2025-09-22","multimodal":false,"param_count":560000000000,"is_new":false},{"rank":8,"model_id":"qwen3-235b-a22b-instruct-2507","model_name":"Qwen3-235B-A22B-Instruct-2507","organization_name":"Alibaba Cloud / Qwen Team","organization_id":"qwen","benchmark_score":0.418,"normalized_score":0.418,"verified":false,"self_reported":true,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"speed_rps":null,"context_window":null,"release_date":"2025-07-22","announcement_date":"2025-07-22","multimodal":false,"param_count":235000000000,"is_new":false}]}