GPT IQ Index
Source: codexradar.comPage updated: Jul 22, 2026, 06:57 AM
GPT IQ Probe
DeepSWE ยท 12 tasksAbove baseline
109.0
Passed
81 / 112
Baseline
75 / 112
08/08 Sampled
Modelgpt-5.6-solDuration60ๅฐๆถ39ๅTokens1.3BCost$1029.71
Experimental metric: a fixed DeepSWE 12-task probe is converted into a GPT IQ index trend; sampling and automation are still being calibrated.
Prediction Radar: 48-hour reset probability
Low probability
24 hours14%48 hours27%
The latest hard reset is complete, while the five-hour limit remains suspended and banked resets provide a separate relief path. Pro usage-drain and client bugs persist, but no prolonged Codex-primary incident or concentrated severe entitlement failure is eviโฆ
ๅฝๅๆฒกๆๅผๅฏ็้่นฌ็ชๅฃExpected window: -
Recent reset windows
No recent window data yet.