{"benchmark_id":"big-bench-extra-hard","benchmark_name":"BIG-Bench Extra Hard","benchmark_description":"BIG-Bench Extra Hard (BBEH) is a challenging benchmark that replaces each task in BIG-Bench Hard with a novel task that probes similar reasoning capabilities but exhibits significantly increased difficulty. The benchmark contains 23 tasks testing diverse reasoning skills including many-hop reasoning, causal understanding, spatial reasoning, temporal arithmetic, geometric reasoning, linguistic reasoning, logic puzzles, and humor understanding. Designed to address saturation on existing benchmarks where state-of-the-art models achieve near-perfect scores, BBEH shows substantial room for improvement with best models achieving only 9.8-44.8% average accuracy.","max_score":1.0,"categories":["language","reasoning","general"],"modality":"text","total_models":11,"entries":[{"rank":1,"model_id":"gemma-4-31b-it","model_name":"Gemma 4 31B","organization_name":"Google","organization_id":"google","benchmark_score":0.744,"normalized_score":0.744,"verified":false,"self_reported":true,"provider_id":"deepinfra","input_cost_per_million":0.13,"output_cost_per_million":0.38,"speed_rps":null,"context_window":262144,"release_date":"2026-04-02","announcement_date":"2026-04-02","multimodal":true,"param_count":30700000000,"is_new":false},{"rank":2,"model_id":"gemma-4-26b-a4b-it","model_name":"Gemma 4 26B-A4B","organization_name":"Google","organization_id":"google","benchmark_score":0.648,"normalized_score":0.648,"verified":false,"self_reported":true,"provider_id":"novita","input_cost_per_million":0.13,"output_cost_per_million":0.4,"speed_rps":null,"context_window":262144,"release_date":"2026-04-02","announcement_date":"2026-04-02","multimodal":true,"param_count":25200000000,"is_new":false},{"rank":3,"model_id":"gemma-4-12b-it","model_name":"Gemma 4 12B","organization_name":"Google","organization_id":"google","benchmark_score":0.53,"normalized_score":0.53,"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":"2026-05-23","announcement_date":"2026-05-23","multimodal":true,"param_count":11959730224,"is_new":false},{"rank":4,"model_id":"diffusiongemma-26b-a4b-it","model_name":"DiffusionGemma 26B-A4B","organization_name":"Google","organization_id":"google","benchmark_score":0.476,"normalized_score":0.476,"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":"2026-06-10","announcement_date":"2026-06-10","multimodal":true,"param_count":25200000000,"is_new":false},{"rank":5,"model_id":"gemma-4-e4b-it","model_name":"Gemma 4 E4B","organization_name":"Google","organization_id":"google","benchmark_score":0.331,"normalized_score":0.331,"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":"2026-04-02","announcement_date":"2026-04-02","multimodal":true,"param_count":8000000000,"is_new":false},{"rank":6,"model_id":"gemma-4-e2b-it","model_name":"Gemma 4 E2B","organization_name":"Google","organization_id":"google","benchmark_score":0.219,"normalized_score":0.219,"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":"2026-04-02","announcement_date":"2026-04-02","multimodal":true,"param_count":5100000000,"is_new":false},{"rank":7,"model_id":"gemma-3-27b-it","model_name":"Gemma 3 27B","organization_name":"Google","organization_id":"google","benchmark_score":0.193,"normalized_score":0.193,"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":"2025-03-12","announcement_date":"2025-03-12","multimodal":true,"param_count":27000000000,"is_new":false},{"rank":8,"model_id":"gemma-3-12b-it","model_name":"Gemma 3 12B","organization_name":"Google","organization_id":"google","benchmark_score":0.163,"normalized_score":0.163,"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":"2025-03-12","announcement_date":"2025-03-12","multimodal":true,"param_count":12000000000,"is_new":false},{"rank":9,"model_id":"gemini-diffusion","model_name":"Gemini Diffusion","organization_name":"Google","organization_id":"google","benchmark_score":0.15,"normalized_score":0.15,"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":"2025-05-20","announcement_date":"2025-05-20","multimodal":false,"param_count":null,"is_new":false},{"rank":10,"model_id":"gemma-3-4b-it","model_name":"Gemma 3 4B","organization_name":"Google","organization_id":"google","benchmark_score":0.11,"normalized_score":0.11,"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":"2025-03-12","announcement_date":"2025-03-12","multimodal":true,"param_count":4000000000,"is_new":false},{"rank":11,"model_id":"gemma-3-1b-it","model_name":"Gemma 3 1B","organization_name":"Google","organization_id":"google","benchmark_score":0.072,"normalized_score":0.072,"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":"2025-03-12","announcement_date":"2025-03-12","multimodal":false,"param_count":1000000000,"is_new":false}]}