{"id":"business-text-20261017","title":"Atendimento e extração de fatos: 12 casos sintéticos, duas rodadas","tested_at":"2026-10-17T09:13:08.046Z","scope":"common-business-long-v1","reference":"OpenAI GPT-4.1 mini via OpenRouter","candidate":"Rota econômica Biziva — texto de negócios","reference_passed":24,"candidate_passed":24,"pairs":24,"reference_usage_usd":0.0375048,"candidate_usage_usd":0.010334160000000004,"retail_formula":"Custo de inferência histórico × multiplicador atual de texto × câmbio declarado; cálculo de venda, não fatura de cliente.","limitations":["Avaliação interna por fatos e JSON; não revisão independente ou garantia de qualidade geral.","Custos observados incluem cache e tokenização reais; não comparação da mesma quantidade de tokens.","Fornecedor intermediário OpenRouter em ambos; não fatura direta da OpenAI.","Não representa retratos, vídeo, áudio, raciocínio geral ou tarefas jurídicas.","O custo de inferência considera o custo acumulado das chamadas pareadas; taxas de compra de crédito, tributos e despesas operacionais não conciliados.","O estado atual da rota depende das aprovações vigentes; este é um resultado histórico."],"results":[{"case_id":"case-0badb29b06101d4935d76a91453ae1bb","round":1,"case_hash":"0badb29b06101d4935d76a91453ae1bbd94704e618dc40a3c63b81b794982b08","reference_pass":1,"candidate_pass":1,"reference_usd":0.0000644,"candidate_usd":0.0000206,"created_at":"2026-10-17T09:24:09.761Z"},{"case_id":"case-0badb29b06101d4935d76a91453ae1bb","round":2,"case_hash":"0badb29b06101d4935d76a91453ae1bbd94704e618dc40a3c63b81b794982b08","reference_pass":1,"candidate_pass":1,"reference_usd":0.0000644,"candidate_usd":0.0000206,"created_at":"2026-10-17T09:24:12.167Z"},{"case_id":"case-1dc5350e86811b1a6f1b55a8ea375d80","round":1,"case_hash":"1dc5350e86811b1a6f1b55a8ea375d80a5f65880e84c8aaf5c7b5e6543618b27","reference_pass":1,"candidate_pass":1,"reference_usd":0.0059136,"candidate_usd":0.001856425,"created_at":"2026-10-17T09:13:08.046Z"},{"case_id":"case-1dc5350e86811b1a6f1b55a8ea375d80","round":2,"case_hash":"1dc5350e86811b1a6f1b55a8ea375d80a5f65880e84c8aaf5c7b5e6543618b27","reference_pass":1,"candidate_pass":1,"reference_usd":0.0016128,"candidate_usd":0.00020077,"created_at":"2026-10-17T09:23:17.818Z"},{"case_id":"case-26b3769ea4b5790c282a57ec924b2962","round":1,"case_hash":"26b3769ea4b5790c282a57ec924b29623b678f36253104d3cdee2f3128faba2b","reference_pass":1,"candidate_pass":1,"reference_usd":0.0013376,"candidate_usd":0.000426425,"created_at":"2026-10-17T09:23:44.602Z"},{"case_id":"case-26b3769ea4b5790c282a57ec924b2962","round":2,"case_hash":"26b3769ea4b5790c282a57ec924b29623b678f36253104d3cdee2f3128faba2b","reference_pass":1,"candidate_pass":1,"reference_usd":0.0004928,"candidate_usd":0.00008637,"created_at":"2026-10-17T09:23:47.090Z"},{"case_id":"case-558f8ca1de61233af56ebe020d816239","round":1,"case_hash":"558f8ca1de61233af56ebe020d816239209b83cc181aee7aba01e6e5ff8e9c08","reference_pass":1,"candidate_pass":1,"reference_usd":0.0023936,"candidate_usd":0.000756425,"created_at":"2026-10-17T09:23:34.979Z"},{"case_id":"case-558f8ca1de61233af56ebe020d816239","round":2,"case_hash":"558f8ca1de61233af56ebe020d816239209b83cc181aee7aba01e6e5ff8e9c08","reference_pass":1,"candidate_pass":1,"reference_usd":0.0007424,"candidate_usd":0.00011277,"created_at":"2026-10-17T09:23:37.375Z"},{"case_id":"case-67c0f8cd547618dedbd7bb428d61b378","round":1,"case_hash":"67c0f8cd547618dedbd7bb428d61b378a36bbbe5600b3fde4ad51eb1ccf3736d","reference_pass":1,"candidate_pass":1,"reference_usd":0.0023952,"candidate_usd":0.000756925,"created_at":"2026-10-17T09:23:39.748Z"},{"case_id":"case-67c0f8cd547618dedbd7bb428d61b378","round":2,"case_hash":"67c0f8cd547618dedbd7bb428d61b378a36bbbe5600b3fde4ad51eb1ccf3736d","reference_pass":1,"candidate_pass":1,"reference_usd":0.000744,"candidate_usd":0.00011327,"created_at":"2026-10-17T09:23:42.143Z"},{"case_id":"case-6b3ad6f247cf5f6cf2ea282cd0cc5e5f","round":1,"case_hash":"6b3ad6f247cf5f6cf2ea282cd0cc5e5f356bf84f82e2d9a45e64c3b92d0d9b54","reference_pass":1,"candidate_pass":1,"reference_usd":0.0013392,"candidate_usd":0.000426925,"created_at":"2026-10-17T09:23:50.559Z"},{"case_id":"case-6b3ad6f247cf5f6cf2ea282cd0cc5e5f","round":2,"case_hash":"6b3ad6f247cf5f6cf2ea282cd0cc5e5f356bf84f82e2d9a45e64c3b92d0d9b54","reference_pass":1,"candidate_pass":1,"reference_usd":0.0004944,"candidate_usd":0.00008687,"created_at":"2026-10-17T09:23:52.882Z"},{"case_id":"case-7ae92340c7c8a93bb77cbf70c3592f2a","round":1,"case_hash":"7ae92340c7c8a93bb77cbf70c3592f2a4f337d871d169054ef1235c7d3ccf717","reference_pass":1,"candidate_pass":1,"reference_usd":0.000068,"candidate_usd":0.0000216,"created_at":"2026-10-17T09:24:04.793Z"},{"case_id":"case-7ae92340c7c8a93bb77cbf70c3592f2a","round":2,"case_hash":"7ae92340c7c8a93bb77cbf70c3592f2a4f337d871d169054ef1235c7d3ccf717","reference_pass":1,"candidate_pass":1,"reference_usd":0.000068,"candidate_usd":0.0000216,"created_at":"2026-10-17T09:24:07.288Z"},{"case_id":"case-b4c183b720d20f13ad17eb94034f4dd3","round":1,"case_hash":"b4c183b720d20f13ad17eb94034f4dd3e870d6db9f754fa03f69e5e75ddc097b","reference_pass":1,"candidate_pass":1,"reference_usd":0.0000904,"candidate_usd":0.000028,"created_at":"2026-10-17T09:24:00.049Z"},{"case_id":"case-b4c183b720d20f13ad17eb94034f4dd3","round":2,"case_hash":"b4c183b720d20f13ad17eb94034f4dd3e870d6db9f754fa03f69e5e75ddc097b","reference_pass":1,"candidate_pass":1,"reference_usd":0.0000904,"candidate_usd":0.000028,"created_at":"2026-10-17T09:24:02.379Z"},{"case_id":"case-b63ca3878bba29f87f450000c9cb094b","round":1,"case_hash":"b63ca3878bba29f87f450000c9cb094b4f9309bf0b902c95e74dfe0650a8ea54","reference_pass":1,"candidate_pass":1,"reference_usd":0.0045936,"candidate_usd":0.001443925,"created_at":"2026-10-17T09:23:25.326Z"},{"case_id":"case-b63ca3878bba29f87f450000c9cb094b","round":2,"case_hash":"b63ca3878bba29f87f450000c9cb094b4f9309bf0b902c95e74dfe0650a8ea54","reference_pass":1,"candidate_pass":1,"reference_usd":0.0012912,"candidate_usd":0.00016777,"created_at":"2026-10-17T09:23:27.714Z"},{"case_id":"case-e25d11e22680eb24eec09382f1e435a4","round":1,"case_hash":"e25d11e22680eb24eec09382f1e435a4e20a23d13a0e588aafcc29e9c0303135","reference_pass":1,"candidate_pass":1,"reference_usd":0.0059152,"candidate_usd":0.001856925,"created_at":"2026-10-17T09:23:20.304Z"},{"case_id":"case-e25d11e22680eb24eec09382f1e435a4","round":2,"case_hash":"e25d11e22680eb24eec09382f1e435a4e20a23d13a0e588aafcc29e9c0303135","reference_pass":1,"candidate_pass":1,"reference_usd":0.0016144,"candidate_usd":0.00020127,"created_at":"2026-10-17T09:23:22.833Z"},{"case_id":"case-f84bedcbbce9db8f0f764d8eaa650d82","round":1,"case_hash":"f84bedcbbce9db8f0f764d8eaa650d82465fbe895202942e6247de9a741d789e","reference_pass":1,"candidate_pass":1,"reference_usd":0.0001456,"candidate_usd":0.000044,"created_at":"2026-10-17T09:23:55.353Z"},{"case_id":"case-f84bedcbbce9db8f0f764d8eaa650d82","round":2,"case_hash":"f84bedcbbce9db8f0f764d8eaa650d82465fbe895202942e6247de9a741d789e","reference_pass":1,"candidate_pass":1,"reference_usd":0.0001456,"candidate_usd":0.000044,"created_at":"2026-10-17T09:23:57.699Z"},{"case_id":"case-fba9cbd2054ddbfb60ad089d3efa320e","round":1,"case_hash":"fba9cbd2054ddbfb60ad089d3efa320e3b0a65605f2ed442cc9dc86b721481b7","reference_pass":1,"candidate_pass":1,"reference_usd":0.0045952,"candidate_usd":0.001444425,"created_at":"2026-10-17T09:23:30.171Z"},{"case_id":"case-fba9cbd2054ddbfb60ad089d3efa320e","round":2,"case_hash":"fba9cbd2054ddbfb60ad089d3efa320e3b0a65605f2ed442cc9dc86b721481b7","reference_pass":1,"candidate_pass":1,"reference_usd":0.0012928,"candidate_usd":0.00016827,"created_at":"2026-10-17T09:23:32.590Z"}],"candidate_model_at_test":"openai/gpt-6-luna","reference_model_at_test":"openai/gpt-4.1-mini","candidate_additional_parameters":{"reasoning":{"effort":"none"}},"retail_not_customer_invoice":true,"quality_review_independent":false,"cases":12,"usage_saving_percent":72.4457669418,"reference_retail_brl":0.4200566666666667,"candidate_retail_brl":0.11574333333333334,"retail_basis":"Custo de inferência histórico × multiplicador atual de texto × câmbio declarado; cálculo de venda, não fatura de cliente.","results_url":"https://api.biziva.com.br/public/benchmark.json","corpus_url":"https://api.biziva.com.br/public/benchmark-corpus.json","reference_invoice":"not_a_direct_openai_invoice","quality_basis":"internal_objective_checks_not_independent_review"}