{"how":"intelligence","basis":"curated","intelligence":5,"category":"general","pick":{"provider":"codex-app-server","model":"gpt-6.1-sol","effort":"medium","fast":false,"run":"codex app-server --stdio · turn/start { model: \"gpt-6.1-sol\", effort: \"medium\" }"},"ladder":{"0":{"provider":"codex-app-server","model":"gpt-6-luna","effort":"low","fast":false,"run":"codex app-server --stdio · turn/start { model: \"gpt-6-luna\", effort: \"low\" }"},"1":{"provider":"codex-app-server","model":"gpt-6-luna","effort":"low","fast":false,"run":"codex app-server --stdio · turn/start { model: \"gpt-6-luna\", effort: \"low\" }"},"2":{"provider":"codex-app-server","model":"gpt-6-luna","effort":"low","fast":false,"run":"codex app-server --stdio · turn/start { model: \"gpt-6-luna\", effort: \"low\" }"},"3":{"provider":"codex-app-server","model":"gpt-6-luna","effort":"low","fast":true,"run":"codex app-server --stdio -c service_tier=\"fast\" · turn/start { model: \"gpt-6-luna\", effort: \"low\" }"},"4":{"provider":"codex-app-server","model":"gpt-6.1-sol","effort":"medium","fast":false,"run":"codex app-server --stdio · turn/start { model: \"gpt-6.1-sol\", effort: \"medium\" }"},"5":{"provider":"codex-app-server","model":"gpt-6.1-sol","effort":"medium","fast":false,"run":"codex app-server --stdio · turn/start { model: \"gpt-6.1-sol\", effort: \"medium\" }"},"6":{"provider":"codex-app-server","model":"gpt-6.1-sol","effort":"medium","fast":false,"run":"codex app-server --stdio · turn/start { model: \"gpt-6.1-sol\", effort: \"medium\" }"},"7":{"provider":"codex-app-server","model":"gpt-6.1-sol","effort":"high","fast":false,"run":"codex app-server --stdio · turn/start { model: \"gpt-6.1-sol\", effort: \"high\" }"},"8":{"provider":"codex-app-server","model":"gpt-6.1-sol","effort":"high","fast":false,"run":"codex app-server --stdio · turn/start { model: \"gpt-6.1-sol\", effort: \"high\" }"},"9":{"provider":"codex-app-server","model":"gpt-6-astra","effort":"high","fast":false,"run":"codex app-server --stdio · turn/start { model: \"gpt-6-astra\", effort: \"high\" }"},"10":{"provider":"codex-app-server","model":"gpt-6-astra","effort":"high","fast":false,"run":"codex app-server --stdio · turn/start { model: \"gpt-6-astra\", effort: \"high\" }"}},"caveats":["Keyword shortlists are curated recommendations, not cross-provider benchmark winners. No Omni head-to-head evaluation has established a universal best model.","Existing benchmark, per-task price, and timing measurements are preserved from the September 2026 import. They were not rerun during the October catalog update.","New model rows have null scores, price, and secs until comparable measurements are captured. Vendor launch benchmarks are not mixed with the older benchmark versions or harnesses.","API token prices are separate from per-task benchmark costs and subscription limits. Long-context, caching, speed tiers, and regional processing can change API pricing.","A signed-in provider does not guarantee access to every model. The installed Codex model cache and Antigravity model list were checked on 2026-10-15; Claude Code account access was not probed with inference.","GPT-5.6 options remain available for explicit selection. GPT-5.5 is retired for ChatGPT-authenticated Codex; API-key sessions are unaffected.","Astra ultra is supported by the local app server but has no separately measured row here; its score is not inferred from max.","Gemini 4 Argon is available on eligible Antigravity plans but absent from the checked account model list, so it is not a default or fallback here."],"providers":["antigravity-cli","claude-code-cli","codex-app-server"],"source":{"benchmark":"Artificial Analysis Intelligence Index v4.3, GDPval-AA v2, Terminal-Bench 4.0","url":"https://artificialanalysis.ai/models/gpt-6-astra","score":"Elo from blind pairwise judging of complete work deliverables over 220 GDPval tasks in an agentic harness, anchored so a human expert scores 1000. 95% CIs run about +/-15 to +/-27, so gaps under ~35 Elo are not separable.","price":"USD per Artificial Analysis Intelligence Index v4.3 task; null means unmeasured.","captured":"2026-09-09","secs":"Seconds per Artificial Analysis Intelligence Index v4.3 task; null means unmeasured.","provider":"Artificial Analysis","design":"Design Arena, Full Stack leaderboard, per-criteria average. Captured 2026-08-25 from https://www.designarena.ai/leaderboard/fullstack","research":"Agents' Last Exam, Score (average partial credit) for the best effort each harness publishes. Captured 2026-08-25 from https://agents-last-exam.org/leaderboard","database":"https://backend.omni.teamofsilicons.com/models.json"},"fresh":true}