Can Agents Run It? · Issue 1: Supabase

Every attempt, on the record

507 execution attempts across two waves. Each one links to the commands the agent actually ran, the output the platform actually returned, and the verifier's decision. Nothing here is summarised — the summaries are on the issue page, and this is what they were made from.

Boundary-axis attempts (d1_edge_fn, d2_oauth) cannot pass by construction and are marked.

By model

ModelPassedAttempts
openai/gpt-5.6-sol3339
openai/gpt-5.6-luna3339
z-ai/glm-5.23239
claude-opus-4-83239
claude-sonnet-53139
google/gemini-3.1-pro-preview3139
moonshotai/kimi-k33139
claude-opus-53139
tencent/hy33139
xiaomi/mimo-v2.53039
deepseek/deepseek-v4-pro2939
minimax/minimax-m32939
qwen/qwen3-coder-plus2239

Every attempt

ModelTaskRepVerdictStepsCostEvidence
claude-opus-4-8d1_auth_email1pass3$0.090open →
claude-opus-4-8d1_auth_email2pass3$0.083open →
claude-opus-4-8d1_auth_email3pass3$0.087open →
claude-opus-5d1_auth_email1pass10$0.354open →
claude-opus-5d1_auth_email2pass8$0.263open →
claude-opus-5d1_auth_email3pass8$0.263open →
claude-sonnet-5d1_auth_email1pass3$0.075open →
claude-sonnet-5d1_auth_email2pass8$0.164open →
claude-sonnet-5d1_auth_email3pass5$0.121open →
deepseek/deepseek-v4-prod1_auth_email1pass7$0.007open →
deepseek/deepseek-v4-prod1_auth_email2pass7$0.006open →
deepseek/deepseek-v4-prod1_auth_email3pass5$0.004open →
google/gemini-3.1-pro-previewd1_auth_email1pass5$0.080open →
google/gemini-3.1-pro-previewd1_auth_email2pass5$0.080open →
google/gemini-3.1-pro-previewd1_auth_email3pass5$0.080open →
minimax/minimax-m3d1_auth_email1pass5$0.008open →
minimax/minimax-m3d1_auth_email2pass8$0.014open →
minimax/minimax-m3d1_auth_email3pass7$0.012open →
moonshotai/kimi-k3d1_auth_email1pass4$0.056open →
moonshotai/kimi-k3d1_auth_email2pass5$0.072open →
moonshotai/kimi-k3d1_auth_email3pass7$0.059open →
openai/gpt-5.6-lunad1_auth_email1pass3$0.008open →
openai/gpt-5.6-lunad1_auth_email2pass3$0.011open →
openai/gpt-5.6-lunad1_auth_email3pass3$0.014open →
openai/gpt-5.6-sold1_auth_email1pass1$0.040open →
openai/gpt-5.6-sold1_auth_email2pass1$0.037open →
openai/gpt-5.6-sold1_auth_email3pass1$0.049open →
qwen/qwen3-coder-plusd1_auth_email1pass9$0.064open →
qwen/qwen3-coder-plusd1_auth_email2pass11$0.105open →
qwen/qwen3-coder-plusd1_auth_email3pass9$0.060open →
tencent/hy3d1_auth_email1pass4$0.003open →
tencent/hy3d1_auth_email2pass4$0.003open →
tencent/hy3d1_auth_email3pass3$0.002open →
xiaomi/mimo-v2.5d1_auth_email1pass4$0.002open →
xiaomi/mimo-v2.5d1_auth_email2pass8$0.005open →
xiaomi/mimo-v2.5d1_auth_email3pass6$0.003open →
z-ai/glm-5.2d1_auth_email1pass5$0.010open →
z-ai/glm-5.2d1_auth_email2pass8$0.022open →
z-ai/glm-5.2d1_auth_email3pass5$0.011open →
claude-opus-4-8d1_crud1pass7$0.255open →
claude-opus-4-8d1_crud2pass5$0.132open →
claude-opus-4-8d1_crud3pass5$0.174open →
claude-opus-5d1_crud1pass15$0.776open →
claude-opus-5d1_crud2pass11$0.392open →
claude-opus-5d1_crud3pass12$0.361open →
claude-sonnet-5d1_crud1pass5$0.084open →
claude-sonnet-5d1_crud2pass5$0.073open →
claude-sonnet-5d1_crud3pass6$0.109open →
deepseek/deepseek-v4-prod1_crud1pass6$0.005open →
deepseek/deepseek-v4-prod1_crud2pass7$0.007open →
deepseek/deepseek-v4-prod1_crud3pass6$0.005open →
google/gemini-3.1-pro-previewd1_crud1pass4$0.044open →
google/gemini-3.1-pro-previewd1_crud2pass3$0.034open →
google/gemini-3.1-pro-previewd1_crud3pass3$0.023open →
minimax/minimax-m3d1_crud1pass6$0.007open →
minimax/minimax-m3d1_crud2pass4$0.005open →
minimax/minimax-m3d1_crud3pass5$0.007open →
moonshotai/kimi-k3d1_crud1pass6$0.118open →
moonshotai/kimi-k3d1_crud2pass3$0.047open →
moonshotai/kimi-k3d1_crud3pass3$0.044open →
openai/gpt-5.6-lunad1_crud1pass3$0.011open →
openai/gpt-5.6-lunad1_crud2pass2$0.009open →
openai/gpt-5.6-lunad1_crud3pass2$0.010open →
openai/gpt-5.6-sold1_crud1pass4$0.080open →
openai/gpt-5.6-sold1_crud2pass4$0.079open →
openai/gpt-5.6-sold1_crud3pass5$0.122open →
qwen/qwen3-coder-plusd1_crud1pass9$0.035open →
qwen/qwen3-coder-plusd1_crud2pass5$0.010open →
qwen/qwen3-coder-plusd1_crud3pass8$0.022open →
tencent/hy3d1_crud1pass3$0.002open →
tencent/hy3d1_crud2pass3$0.002open →
tencent/hy3d1_crud3pass4$0.002open →
xiaomi/mimo-v2.5d1_crud1pass3$0.001open →
xiaomi/mimo-v2.5d1_crud2pass4$0.001open →
xiaomi/mimo-v2.5d1_crud3pass4$0.002open →
z-ai/glm-5.2d1_crud1pass4$0.011open →
z-ai/glm-5.2d1_crud2pass5$0.011open →
z-ai/glm-5.2d1_crud3pass5$0.010open →
claude-opus-4-8d1_edge_fn ⟂1fail16$0.998open →
claude-opus-4-8d1_edge_fn ⟂2fail17$0.965open →
claude-opus-4-8d1_edge_fn ⟂3fail18$1.028open →
claude-opus-5d1_edge_fn ⟂1fail20$1.021open →
claude-opus-5d1_edge_fn ⟂2fail24$1.033open →
claude-opus-5d1_edge_fn ⟂3fail12$0.277open →
claude-sonnet-5d1_edge_fn ⟂1fail15$0.486open →
claude-sonnet-5d1_edge_fn ⟂2fail14$0.522open →
claude-sonnet-5d1_edge_fn ⟂3fail15$0.489open →
deepseek/deepseek-v4-prod1_edge_fn ⟂1fail20$0.050open →
deepseek/deepseek-v4-prod1_edge_fn ⟂2fail22$0.050open →
deepseek/deepseek-v4-prod1_edge_fn ⟂3fail21$0.049open →
google/gemini-3.1-pro-previewd1_edge_fn ⟂1fail24$0.502open →
google/gemini-3.1-pro-previewd1_edge_fn ⟂2fail24$0.497open →
google/gemini-3.1-pro-previewd1_edge_fn ⟂3fail24$0.489open →
minimax/minimax-m3d1_edge_fn ⟂1fail23$0.058open →
minimax/minimax-m3d1_edge_fn ⟂2fail21$0.059open →
minimax/minimax-m3d1_edge_fn ⟂3fail23$0.059open →
moonshotai/kimi-k3d1_edge_fn ⟂1fail31$0.634open →
moonshotai/kimi-k3d1_edge_fn ⟂2fail31$0.641open →
moonshotai/kimi-k3d1_edge_fn ⟂3fail26$0.663open →
openai/gpt-5.6-lunad1_edge_fn ⟂1fail11$0.076open →
openai/gpt-5.6-lunad1_edge_fn ⟂2fail22$0.244open →
openai/gpt-5.6-lunad1_edge_fn ⟂3fail16$0.187open →
openai/gpt-5.6-sold1_edge_fn ⟂1fail22$0.853open →
openai/gpt-5.6-sold1_edge_fn ⟂2fail20$1.169open →
openai/gpt-5.6-sold1_edge_fn ⟂3fail20$1.220open →
qwen/qwen3-coder-plusd1_edge_fn ⟂1fail19$0.139open →
qwen/qwen3-coder-plusd1_edge_fn ⟂2fail13$0.065open →
qwen/qwen3-coder-plusd1_edge_fn ⟂3fail20$0.138open →
tencent/hy3d1_edge_fn ⟂1fail19$0.026open →
tencent/hy3d1_edge_fn ⟂2fail17$0.025open →
tencent/hy3d1_edge_fn ⟂3fail15$0.024open →
xiaomi/mimo-v2.5d1_edge_fn ⟂1fail18$0.017open →
xiaomi/mimo-v2.5d1_edge_fn ⟂2fail18$0.017open →
xiaomi/mimo-v2.5d1_edge_fn ⟂3fail17$0.016open →
z-ai/glm-5.2d1_edge_fn ⟂1fail21$0.124open →
z-ai/glm-5.2d1_edge_fn ⟂2fail22$0.131open →
z-ai/glm-5.2d1_edge_fn ⟂3fail28$0.127open →
claude-opus-4-8d1_rls_owner1pass6$0.319open →
claude-opus-4-8d1_rls_owner2pass10$0.534open →
claude-opus-4-8d1_rls_owner3pass10$0.539open →
claude-opus-5d1_rls_owner1pass18$0.890open →
claude-opus-5d1_rls_owner2pass19$0.971open →
claude-opus-5d1_rls_owner3pass15$0.830open →
claude-sonnet-5d1_rls_owner1fail11$0.558open →
claude-sonnet-5d1_rls_owner2pass10$0.351open →
claude-sonnet-5d1_rls_owner3fail8$0.292open →
deepseek/deepseek-v4-prod1_rls_owner1fail20$0.051open →
deepseek/deepseek-v4-prod1_rls_owner2fail4$0.003open →
deepseek/deepseek-v4-prod1_rls_owner3pass20$0.049open →
google/gemini-3.1-pro-previewd1_rls_owner1pass15$0.202open →
google/gemini-3.1-pro-previewd1_rls_owner2pass14$0.277open →
google/gemini-3.1-pro-previewd1_rls_owner3pass20$0.528open →
minimax/minimax-m3d1_rls_owner1pass12$0.023open →
minimax/minimax-m3d1_rls_owner2pass8$0.016open →
minimax/minimax-m3d1_rls_owner3pass16$0.035open →
moonshotai/kimi-k3d1_rls_owner1pass9$0.153open →
moonshotai/kimi-k3d1_rls_owner2pass11$0.243open →
moonshotai/kimi-k3d1_rls_owner3pass11$0.230open →
openai/gpt-5.6-lunad1_rls_owner1pass7$0.029open →
openai/gpt-5.6-lunad1_rls_owner2pass7$0.030open →
openai/gpt-5.6-lunad1_rls_owner3pass14$0.045open →
openai/gpt-5.6-sold1_rls_owner1pass5$0.148open →
openai/gpt-5.6-sold1_rls_owner2pass5$0.129open →
openai/gpt-5.6-sold1_rls_owner3pass4$0.093open →
qwen/qwen3-coder-plusd1_rls_owner1pass10$0.033open →
qwen/qwen3-coder-plusd1_rls_owner2pass12$0.045open →
qwen/qwen3-coder-plusd1_rls_owner3pass9$0.036open →
tencent/hy3d1_rls_owner1pass8$0.011open →
tencent/hy3d1_rls_owner2pass11$0.013open →
tencent/hy3d1_rls_owner3pass16$0.017open →
xiaomi/mimo-v2.5d1_rls_owner1pass15$0.011open →
xiaomi/mimo-v2.5d1_rls_owner2pass19$0.014open →
xiaomi/mimo-v2.5d1_rls_owner3pass14$0.009open →
z-ai/glm-5.2d1_rls_owner1pass20$0.078open →
z-ai/glm-5.2d1_rls_owner2pass19$0.120open →
z-ai/glm-5.2d1_rls_owner3pass22$0.122open →
claude-opus-4-8d1_schema1pass6$0.218open →
claude-opus-4-8d1_schema2pass6$0.222open →
claude-opus-4-8d1_schema3pass6$0.213open →
claude-opus-5d1_schema1pass19$0.940open →
claude-opus-5d1_schema2pass15$0.644open →
claude-opus-5d1_schema3pass14$0.541open →
claude-sonnet-5d1_schema1pass4$0.056open →
claude-sonnet-5d1_schema2pass5$0.082open →
claude-sonnet-5d1_schema3pass5$0.101open →
deepseek/deepseek-v4-prod1_schema1pass6$0.009open →
deepseek/deepseek-v4-prod1_schema2pass6$0.008open →
deepseek/deepseek-v4-prod1_schema3pass4$0.005open →
google/gemini-3.1-pro-previewd1_schema1pass4$0.058open →
google/gemini-3.1-pro-previewd1_schema2pass3$0.043open →
google/gemini-3.1-pro-previewd1_schema3pass3$0.040open →
minimax/minimax-m3d1_schema1pass12$0.023open →
minimax/minimax-m3d1_schema2pass8$0.020open →
minimax/minimax-m3d1_schema3pass21$0.060open →
moonshotai/kimi-k3d1_schema1pass7$0.137open →
moonshotai/kimi-k3d1_schema2pass7$0.156open →
moonshotai/kimi-k3d1_schema3pass5$0.054open →
openai/gpt-5.6-lunad1_schema1pass4$0.014open →
openai/gpt-5.6-lunad1_schema2pass4$0.016open →
openai/gpt-5.6-lunad1_schema3pass4$0.013open →
openai/gpt-5.6-sold1_schema1pass4$0.095open →
openai/gpt-5.6-sold1_schema2pass4$0.082open →
openai/gpt-5.6-sold1_schema3pass4$0.079open →
qwen/qwen3-coder-plusd1_schema1pass6$0.018open →
qwen/qwen3-coder-plusd1_schema2pass8$0.031open →
qwen/qwen3-coder-plusd1_schema3pass9$0.042open →
tencent/hy3d1_schema1pass5$0.003open →
tencent/hy3d1_schema2pass6$0.005open →
tencent/hy3d1_schema3pass6$0.005open →
xiaomi/mimo-v2.5d1_schema1pass7$0.003open →
xiaomi/mimo-v2.5d1_schema2pass6$0.002open →
xiaomi/mimo-v2.5d1_schema3pass4$0.002open →
z-ai/glm-5.2d1_schema1pass8$0.023open →
z-ai/glm-5.2d1_schema2pass16$0.065open →
z-ai/glm-5.2d1_schema3pass8$0.021open →
claude-opus-4-8d1_storage1pass7$0.229open →
claude-opus-4-8d1_storage2pass7$0.231open →
claude-opus-4-8d1_storage3pass6$0.227open →
claude-opus-5d1_storage1pass12$0.364open →
claude-opus-5d1_storage2pass15$0.406open →
claude-opus-5d1_storage3pass17$0.502open →
claude-sonnet-5d1_storage1pass8$0.183open →
claude-sonnet-5d1_storage2pass6$0.163open →
claude-sonnet-5d1_storage3pass9$0.252open →
deepseek/deepseek-v4-prod1_storage1pass6$0.007open →
deepseek/deepseek-v4-prod1_storage2pass8$0.005open →
deepseek/deepseek-v4-prod1_storage3pass8$0.005open →
google/gemini-3.1-pro-previewd1_storage1pass9$0.083open →
google/gemini-3.1-pro-previewd1_storage2pass10$0.185open →
google/gemini-3.1-pro-previewd1_storage3pass6$0.067open →
minimax/minimax-m3d1_storage1pass7$0.013open →
minimax/minimax-m3d1_storage2pass10$0.017open →
minimax/minimax-m3d1_storage3pass5$0.009open →
moonshotai/kimi-k3d1_storage1pass9$0.082open →
moonshotai/kimi-k3d1_storage2pass5$0.063open →
moonshotai/kimi-k3d1_storage3pass12$0.138open →
openai/gpt-5.6-lunad1_storage1pass3$0.013open →
openai/gpt-5.6-lunad1_storage2pass3$0.013open →
openai/gpt-5.6-lunad1_storage3pass2$0.009open →
openai/gpt-5.6-sold1_storage1pass2$0.058open →
openai/gpt-5.6-sold1_storage2pass3$0.086open →
openai/gpt-5.6-sold1_storage3pass2$0.061open →
qwen/qwen3-coder-plusd1_storage1pass11$0.066open →
qwen/qwen3-coder-plusd1_storage2pass14$0.046open →
qwen/qwen3-coder-plusd1_storage3pass17$0.136open →
tencent/hy3d1_storage1pass5$0.003open →
tencent/hy3d1_storage2pass5$0.004open →
tencent/hy3d1_storage3pass6$0.004open →
xiaomi/mimo-v2.5d1_storage1pass8$0.002open →
xiaomi/mimo-v2.5d1_storage2pass9$0.002open →
xiaomi/mimo-v2.5d1_storage3pass8$0.003open →
z-ai/glm-5.2d1_storage1pass9$0.021open →
z-ai/glm-5.2d1_storage2pass12$0.020open →
z-ai/glm-5.2d1_storage3pass13$0.037open →
claude-opus-4-8d2_backup_restore1pass10$0.468open →
claude-opus-4-8d2_backup_restore2pass11$0.446open →
claude-opus-4-8d2_backup_restore3pass11$0.572open →
claude-opus-5d2_backup_restore1pass25$1.022open →
claude-opus-5d2_backup_restore2pass19$0.973open →
claude-opus-5d2_backup_restore3pass22$1.022open →
claude-sonnet-5d2_backup_restore1pass14$0.479open →
claude-sonnet-5d2_backup_restore2pass9$0.352open →
claude-sonnet-5d2_backup_restore3pass13$0.304open →
deepseek/deepseek-v4-prod2_backup_restore1pass8$0.011open →
deepseek/deepseek-v4-prod2_backup_restore2pass14$0.028open →
deepseek/deepseek-v4-prod2_backup_restore3pass13$0.016open →
google/gemini-3.1-pro-previewd2_backup_restore1pass12$0.128open →
google/gemini-3.1-pro-previewd2_backup_restore2pass14$0.155open →
google/gemini-3.1-pro-previewd2_backup_restore3pass15$0.159open →
minimax/minimax-m3d2_backup_restore1pass25$0.033open →
minimax/minimax-m3d2_backup_restore2pass8$0.014open →
minimax/minimax-m3d2_backup_restore3pass36$0.057open →
moonshotai/kimi-k3d2_backup_restore1pass8$0.151open →
moonshotai/kimi-k3d2_backup_restore2fail25$0.606open →
moonshotai/kimi-k3d2_backup_restore3pass8$0.159open →
openai/gpt-5.6-lunad2_backup_restore1pass3$0.014open →
openai/gpt-5.6-lunad2_backup_restore2pass7$0.039open →
openai/gpt-5.6-lunad2_backup_restore3pass5$0.023open →
openai/gpt-5.6-sold2_backup_restore1pass5$0.163open →
openai/gpt-5.6-sold2_backup_restore2pass5$0.159open →
openai/gpt-5.6-sold2_backup_restore3pass6$0.252open →
qwen/qwen3-coder-plusd2_backup_restore1fail15$0.145open →
qwen/qwen3-coder-plusd2_backup_restore2pass17$0.139open →
qwen/qwen3-coder-plusd2_backup_restore3fail19$0.132open →
tencent/hy3d2_backup_restore1pass10$0.011open →
tencent/hy3d2_backup_restore2pass9$0.008open →
tencent/hy3d2_backup_restore3pass9$0.009open →
xiaomi/mimo-v2.5d2_backup_restore1pass13$0.009open →
xiaomi/mimo-v2.5d2_backup_restore2pass10$0.005open →
xiaomi/mimo-v2.5d2_backup_restore3pass8$0.004open →
z-ai/glm-5.2d2_backup_restore1pass7$0.026open →
z-ai/glm-5.2d2_backup_restore2pass8$0.029open →
z-ai/glm-5.2d2_backup_restore3pass7$0.036open →
claude-opus-4-8d2_live_migration1pass12$0.514open →
claude-opus-4-8d2_live_migration2pass8$0.231open →
claude-opus-4-8d2_live_migration3pass9$0.320open →
claude-opus-5d2_live_migration1pass16$0.559open →
claude-opus-5d2_live_migration2pass12$0.483open →
claude-opus-5d2_live_migration3pass20$0.919open →
claude-sonnet-5d2_live_migration1pass7$0.153open →
claude-sonnet-5d2_live_migration2pass5$0.118open →
claude-sonnet-5d2_live_migration3pass6$0.161open →
deepseek/deepseek-v4-prod2_live_migration1pass8$0.013open →
deepseek/deepseek-v4-prod2_live_migration2pass13$0.021open →
deepseek/deepseek-v4-prod2_live_migration3pass8$0.013open →
google/gemini-3.1-pro-previewd2_live_migration1pass7$0.059open →
google/gemini-3.1-pro-previewd2_live_migration2pass6$0.055open →
google/gemini-3.1-pro-previewd2_live_migration3pass7$0.074open →
minimax/minimax-m3d2_live_migration1pass9$0.017open →
minimax/minimax-m3d2_live_migration2pass11$0.015open →
minimax/minimax-m3d2_live_migration3pass6$0.012open →
moonshotai/kimi-k3d2_live_migration1pass11$0.225open →
moonshotai/kimi-k3d2_live_migration2pass7$0.114open →
moonshotai/kimi-k3d2_live_migration3pass8$0.139open →
openai/gpt-5.6-lunad2_live_migration1pass9$0.022open →
openai/gpt-5.6-lunad2_live_migration2pass3$0.016open →
openai/gpt-5.6-lunad2_live_migration3pass3$0.016open →
openai/gpt-5.6-sold2_live_migration1pass3$0.068open →
openai/gpt-5.6-sold2_live_migration2pass4$0.082open →
openai/gpt-5.6-sold2_live_migration3pass6$0.120open →
qwen/qwen3-coder-plusd2_live_migration1fail4$0.006open →
qwen/qwen3-coder-plusd2_live_migration2pass15$0.059open →
qwen/qwen3-coder-plusd2_live_migration3pass14$0.044open →
tencent/hy3d2_live_migration1pass7$0.006open →
tencent/hy3d2_live_migration2pass5$0.004open →
tencent/hy3d2_live_migration3pass7$0.005open →
xiaomi/mimo-v2.5d2_live_migration1pass5$0.002open →
xiaomi/mimo-v2.5d2_live_migration2pass15$0.009open →
xiaomi/mimo-v2.5d2_live_migration3pass12$0.003open →
z-ai/glm-5.2d2_live_migration1pass7$0.021open →
z-ai/glm-5.2d2_live_migration2pass5$0.013open →
z-ai/glm-5.2d2_live_migration3pass7$0.021open →
claude-opus-4-8d2_oauth ⟂1fail18$0.970open →
claude-opus-4-8d2_oauth ⟂2fail15$0.964open →
claude-opus-4-8d2_oauth ⟂3fail15$1.024open →
claude-opus-5d2_oauth ⟂1fail8$0.149open →
claude-opus-5d2_oauth ⟂2fail8$0.190open →
claude-opus-5d2_oauth ⟂3fail12$0.361open →
claude-sonnet-5d2_oauth ⟂1fail13$0.498open →
claude-sonnet-5d2_oauth ⟂2fail15$0.502open →
claude-sonnet-5d2_oauth ⟂3fail16$0.501open →
deepseek/deepseek-v4-prod2_oauth ⟂1fail21$0.052open →
deepseek/deepseek-v4-prod2_oauth ⟂2fail20$0.048open →
deepseek/deepseek-v4-prod2_oauth ⟂3fail24$0.050open →
google/gemini-3.1-pro-previewd2_oauth ⟂1fail22$0.503open →
google/gemini-3.1-pro-previewd2_oauth ⟂2fail21$0.491open →
google/gemini-3.1-pro-previewd2_oauth ⟂3fail21$0.497open →
minimax/minimax-m3d2_oauth ⟂1fail18$0.058open →
minimax/minimax-m3d2_oauth ⟂2fail22$0.057open →
minimax/minimax-m3d2_oauth ⟂3fail19$0.059open →
moonshotai/kimi-k3d2_oauth ⟂1fail27$0.602open →
moonshotai/kimi-k3d2_oauth ⟂2fail25$0.642open →
moonshotai/kimi-k3d2_oauth ⟂3fail25$0.602open →
openai/gpt-5.6-lunad2_oauth ⟂1fail27$0.239open →
openai/gpt-5.6-lunad2_oauth ⟂2fail21$0.243open →
openai/gpt-5.6-lunad2_oauth ⟂3fail19$0.240open →
openai/gpt-5.6-sold2_oauth ⟂1fail26$1.181open →
openai/gpt-5.6-sold2_oauth ⟂2fail30$1.159open →
openai/gpt-5.6-sold2_oauth ⟂3fail37$1.253open →
qwen/qwen3-coder-plusd2_oauth ⟂1fail19$0.134open →
qwen/qwen3-coder-plusd2_oauth ⟂2fail25$0.137open →
qwen/qwen3-coder-plusd2_oauth ⟂3fail22$0.139open →
tencent/hy3d2_oauth ⟂1fail26$0.026open →
tencent/hy3d2_oauth ⟂2fail16$0.025open →
tencent/hy3d2_oauth ⟂3fail14$0.025open →
xiaomi/mimo-v2.5d2_oauth ⟂1fail17$0.017open →
xiaomi/mimo-v2.5d2_oauth ⟂2fail20$0.016open →
xiaomi/mimo-v2.5d2_oauth ⟂3fail23$0.018open →
z-ai/glm-5.2d2_oauth ⟂1fail25$0.122open →
z-ai/glm-5.2d2_oauth ⟂2fail36$0.131open →
z-ai/glm-5.2d2_oauth ⟂3fail29$0.127open →
claude-opus-4-8d2_pgvector1pass6$0.207open →
claude-opus-4-8d2_pgvector2pass8$0.293open →
claude-opus-4-8d2_pgvector3pass5$0.159open →
claude-opus-5d2_pgvector1pass19$0.871open →
claude-opus-5d2_pgvector2pass20$0.885open →
claude-opus-5d2_pgvector3pass20$0.922open →
claude-sonnet-5d2_pgvector1pass9$0.202open →
claude-sonnet-5d2_pgvector2pass9$0.216open →
claude-sonnet-5d2_pgvector3pass7$0.166open →
deepseek/deepseek-v4-prod2_pgvector1pass12$0.013open →
deepseek/deepseek-v4-prod2_pgvector2pass10$0.011open →
deepseek/deepseek-v4-prod2_pgvector3pass11$0.013open →
google/gemini-3.1-pro-previewd2_pgvector1pass5$0.054open →
google/gemini-3.1-pro-previewd2_pgvector2pass7$0.078open →
google/gemini-3.1-pro-previewd2_pgvector3pass5$0.055open →
minimax/minimax-m3d2_pgvector1pass9$0.014open →
minimax/minimax-m3d2_pgvector2pass18$0.021open →
minimax/minimax-m3d2_pgvector3pass6$0.008open →
moonshotai/kimi-k3d2_pgvector1pass9$0.185open →
moonshotai/kimi-k3d2_pgvector2pass6$0.110open →
moonshotai/kimi-k3d2_pgvector3pass8$0.154open →
openai/gpt-5.6-lunad2_pgvector1pass4$0.016open →
openai/gpt-5.6-lunad2_pgvector2pass3$0.013open →
openai/gpt-5.6-lunad2_pgvector3pass4$0.017open →
openai/gpt-5.6-sold2_pgvector1pass6$0.119open →
openai/gpt-5.6-sold2_pgvector2pass4$0.094open →
openai/gpt-5.6-sold2_pgvector3pass3$0.075open →
qwen/qwen3-coder-plusd2_pgvector1fail10$0.030open →
qwen/qwen3-coder-plusd2_pgvector2fail7$0.015open →
qwen/qwen3-coder-plusd2_pgvector3fail10$0.024open →
tencent/hy3d2_pgvector1pass15$0.015open →
tencent/hy3d2_pgvector2pass11$0.010open →
tencent/hy3d2_pgvector3pass10$0.008open →
xiaomi/mimo-v2.5d2_pgvector1pass14$0.007open →
xiaomi/mimo-v2.5d2_pgvector2pass14$0.007open →
xiaomi/mimo-v2.5d2_pgvector3pass10$0.004open →
z-ai/glm-5.2d2_pgvector1pass10$0.035open →
z-ai/glm-5.2d2_pgvector2pass8$0.022open →
z-ai/glm-5.2d2_pgvector3pass12$0.028open →
claude-opus-4-8d2_pooling1pass8$0.391open →
claude-opus-4-8d2_pooling2pass7$0.313open →
claude-opus-4-8d2_pooling3pass8$0.391open →
claude-opus-5d2_pooling1fail18$1.038open →
claude-opus-5d2_pooling2pass19$1.060open →
claude-opus-5d2_pooling3pass19$1.032open →
claude-sonnet-5d2_pooling1pass9$0.218open →
claude-sonnet-5d2_pooling2pass11$0.379open →
claude-sonnet-5d2_pooling3pass7$0.193open →
deepseek/deepseek-v4-prod2_pooling1pass6$0.013open →
deepseek/deepseek-v4-prod2_pooling2pass8$0.009open →
deepseek/deepseek-v4-prod2_pooling3pass7$0.008open →
google/gemini-3.1-pro-previewd2_pooling1pass7$0.103open →
google/gemini-3.1-pro-previewd2_pooling2pass6$0.093open →
google/gemini-3.1-pro-previewd2_pooling3pass5$0.077open →
minimax/minimax-m3d2_pooling1pass5$0.010open →
minimax/minimax-m3d2_pooling2pass7$0.017open →
minimax/minimax-m3d2_pooling3pass8$0.017open →
moonshotai/kimi-k3d2_pooling1fail9$0.146open →
moonshotai/kimi-k3d2_pooling2pass10$0.211open →
moonshotai/kimi-k3d2_pooling3pass9$0.149open →
openai/gpt-5.6-lunad2_pooling1pass2$0.011open →
openai/gpt-5.6-lunad2_pooling2pass3$0.014open →
openai/gpt-5.6-lunad2_pooling3pass4$0.028open →
openai/gpt-5.6-sold2_pooling1pass3$0.055open →
openai/gpt-5.6-sold2_pooling2pass3$0.055open →
openai/gpt-5.6-sold2_pooling3pass2$0.054open →
qwen/qwen3-coder-plusd2_pooling1fail15$0.132open →
qwen/qwen3-coder-plusd2_pooling2pass4$0.013open →
qwen/qwen3-coder-plusd2_pooling3fail18$0.144open →
tencent/hy3d2_pooling1pass7$0.008open →
tencent/hy3d2_pooling2fail0$0.001open →
tencent/hy3d2_pooling3pass6$0.005open →
xiaomi/mimo-v2.5d2_pooling1fail6$0.004open →
xiaomi/mimo-v2.5d2_pooling2pass7$0.003open →
xiaomi/mimo-v2.5d2_pooling3pass9$0.005open →
z-ai/glm-5.2d2_pooling1pass8$0.029open →
z-ai/glm-5.2d2_pooling2pass9$0.025open →
z-ai/glm-5.2d2_pooling3fail28$0.126open →
claude-opus-4-8d2_rls_tenant1pass10$0.645open →
claude-opus-4-8d2_rls_tenant2fail14$1.021open →
claude-opus-4-8d2_rls_tenant3pass13$0.990open →
claude-opus-5d2_rls_tenant1pass15$0.987open →
claude-opus-5d2_rls_tenant2fail16$0.971open →
claude-opus-5d2_rls_tenant3pass16$1.065open →
claude-sonnet-5d2_rls_tenant1pass14$0.516open →
claude-sonnet-5d2_rls_tenant2pass14$0.496open →
claude-sonnet-5d2_rls_tenant3pass9$0.397open →
deepseek/deepseek-v4-prod2_rls_tenant1pass14$0.051open →
deepseek/deepseek-v4-prod2_rls_tenant2fail23$0.051open →
deepseek/deepseek-v4-prod2_rls_tenant3pass20$0.050open →
google/gemini-3.1-pro-previewd2_rls_tenant1fail22$0.515open →
google/gemini-3.1-pro-previewd2_rls_tenant2pass16$0.520open →
google/gemini-3.1-pro-previewd2_rls_tenant3fail16$0.286open →
minimax/minimax-m3d2_rls_tenant1fail16$0.056open →
minimax/minimax-m3d2_rls_tenant2pass11$0.045open →
minimax/minimax-m3d2_rls_tenant3pass14$0.057open →
moonshotai/kimi-k3d2_rls_tenant1pass11$0.320open →
moonshotai/kimi-k3d2_rls_tenant2pass11$0.451open →
moonshotai/kimi-k3d2_rls_tenant3pass13$0.401open →
openai/gpt-5.6-lunad2_rls_tenant1pass12$0.090open →
openai/gpt-5.6-lunad2_rls_tenant2pass8$0.063open →
openai/gpt-5.6-lunad2_rls_tenant3pass8$0.064open →
openai/gpt-5.6-sold2_rls_tenant1pass7$0.369open →
openai/gpt-5.6-sold2_rls_tenant2pass3$0.127open →
openai/gpt-5.6-sold2_rls_tenant3pass5$0.202open →
qwen/qwen3-coder-plusd2_rls_tenant1fail12$0.048open →
qwen/qwen3-coder-plusd2_rls_tenant2fail8$0.036open →
qwen/qwen3-coder-plusd2_rls_tenant3fail11$0.045open →
tencent/hy3d2_rls_tenant1pass17$0.027open →
tencent/hy3d2_rls_tenant2pass13$0.023open →
tencent/hy3d2_rls_tenant3pass14$0.025open →
xiaomi/mimo-v2.5d2_rls_tenant1fail17$0.018open →
xiaomi/mimo-v2.5d2_rls_tenant2fail15$0.016open →
xiaomi/mimo-v2.5d2_rls_tenant3pass13$0.017open →
z-ai/glm-5.2d2_rls_tenant1pass19$0.126open →
z-ai/glm-5.2d2_rls_tenant2pass12$0.043open →
z-ai/glm-5.2d2_rls_tenant3pass19$0.123open →
claude-opus-4-8d2_webhooks1pass13$0.611open →
claude-opus-4-8d2_webhooks2pass16$0.650open →
claude-opus-4-8d2_webhooks3pass15$0.795open →
claude-opus-5d2_webhooks1pass21$1.001open →
claude-opus-5d2_webhooks2pass21$1.015open →
claude-opus-5d2_webhooks3pass20$0.977open →
claude-sonnet-5d2_webhooks1pass18$0.525open →
claude-sonnet-5d2_webhooks2pass15$0.402open →
claude-sonnet-5d2_webhooks3pass15$0.442open →
deepseek/deepseek-v4-prod2_webhooks1pass24$0.051open →
deepseek/deepseek-v4-prod2_webhooks2fail23$0.050open →
deepseek/deepseek-v4-prod2_webhooks3pass21$0.048open →
google/gemini-3.1-pro-previewd2_webhooks1pass31$0.485open →
google/gemini-3.1-pro-previewd2_webhooks2pass15$0.210open →
google/gemini-3.1-pro-previewd2_webhooks3pass23$0.267open →
minimax/minimax-m3d2_webhooks1fail22$0.057open →
minimax/minimax-m3d2_webhooks2fail21$0.058open →
minimax/minimax-m3d2_webhooks3fail40$0.050open →
moonshotai/kimi-k3d2_webhooks1pass10$0.211open →
moonshotai/kimi-k3d2_webhooks2pass8$0.162open →
moonshotai/kimi-k3d2_webhooks3pass9$0.204open →
openai/gpt-5.6-lunad2_webhooks1pass6$0.036open →
openai/gpt-5.6-lunad2_webhooks2pass8$0.049open →
openai/gpt-5.6-lunad2_webhooks3pass17$0.088open →
openai/gpt-5.6-sold2_webhooks1pass10$0.259open →
openai/gpt-5.6-sold2_webhooks2pass8$0.359open →
openai/gpt-5.6-sold2_webhooks3pass6$0.210open →
qwen/qwen3-coder-plusd2_webhooks1pass23$0.131open →
qwen/qwen3-coder-plusd2_webhooks2pass18$0.080open →
qwen/qwen3-coder-plusd2_webhooks3pass25$0.130open →
tencent/hy3d2_webhooks1pass13$0.023open →
tencent/hy3d2_webhooks2fail5$0.004open →
tencent/hy3d2_webhooks3pass11$0.012open →
xiaomi/mimo-v2.5d2_webhooks1pass18$0.011open →
xiaomi/mimo-v2.5d2_webhooks2pass21$0.016open →
xiaomi/mimo-v2.5d2_webhooks3pass20$0.013open →
z-ai/glm-5.2d2_webhooks1pass25$0.132open →
z-ai/glm-5.2d2_webhooks2pass23$0.054open →
z-ai/glm-5.2d2_webhooks3pass23$0.052open →

Source data, task definitions and verifier scripts: the repository.