Codex

Codex Radar

designed by Codex
中文
Codex Notes

How to enable Max reasoning effort

Quota Radar 7月16日09:47更新

The 5h limit is temporarily paused, so Quota Radar currently shows only the active 7d quota. GPT-5.6 estimates include an estimated cache-write surcharge.

Tier 7d Basis
20x Pro $1,428.41 measured 7d
5x Pro $357.10 estimated
Plus $71.42 estimated
Quota change trend $1,922.96 → $1,428.41 (-$494.55, -25.7%)
$1,315 $1,742 $2,169 2026-07-09-pm 20x Pro 7d $1,461.73 2026-07-11-pm 20x Pro 7d $2,034.52 2026-07-12-am 20x Pro 7d $1,751.38 2026-07-12-pm 20x Pro 7d $2,055.86 2026-07-13-pm 20x Pro 7d $1,866.08 2026-07-14-am 20x Pro 7d $1,573.78 2026-07-14-pm 20x Pro 7d $1,722.06 2026-07-15-am 20x Pro 7d $1,583.74 2026-07-15-pm 20x Pro 7d $1,922.96 2026-07-16-am 20x Pro 7d $1,428.41 7.9_pm 7.11_pm 7.12_am 7.12_pm 7.13_pm 7.14_am 7.14_pm 7.15_am 7.15_pm 7.16_am
Method

Using the latest valid 7d calibration (2026-07-16-am) from a 20x Pro US account. The 5h limit is temporarily paused and is not displayed. Formula: 20x Pro 7d = $214.26 / 15% * 100 = $1,428.41; the API-equivalent cost includes the estimated cache-write surcharge.

Fast Radar Updated Jul 14, 18:01

Switching Standard to Fast: how much faster for 2.5× the cost?
Perceived acceleration⚡️1.357×
First-token latency saved2.19s
Token generation acceleration⚡️1.477×
Model Perceived acceleration · E2E First-token latency reduction · TTFT Token generation acceleration · TPS
Sol
56.29s → 40.35s⚡️1.395×
19.17s → 15.53s19.0% faster
55.93 → 83.75⚡️1.498×
Terra
59.64s → 41.87s⚡️1.425×
18.14s → 13.92s23.2% faster
50.74 → 75.40⚡️1.486×
Luna
50.68s → 40.56s⚡️1.250×
13.19s → 14.47s9.7% slower
55.43 → 80.21⚡️1.447×
Method

Method: Sol, Terra, and Luna use low reasoning effort with the same count-from-1-to-1024 output task. Standard and Fast each run three independent samples and use the arithmetic mean; order alternates, up to three model pairs run concurrently, and each pair still runs Standard/Fast serially. Every sample uses a fresh app-server and ephemeral thread plus a random nonce. Complete outputs with cached_input_tokens = 0 are preferred. For this Jul 14 run, slow samples and one complete Terra cache-mismatched pair are retained as real same-day experience; incomplete outputs are rejected and rerun. TTFT is delay to the first visible token, TPS is sustained token generation speed, and E2E is total time from submission to the complete result. Headline acceleration values are calculated as mean Standard divided by mean Fast; per-model E2E acceleration uses that model's Standard divided by Fast.

IQ Drop Radar Updated Jul 16, 09:48

Updated twice daily, around 7:00-9:00 and 13:00-15:00 Beijing time.

This Codex multi-model IQ run consumed an estimated $214.26 in API-equivalent cost. Click model cards to filter which models are shown.

Sol max
105.0$60.24.9h
Sol xhigh
105.0$35.62.9h
Sol high
105.0$23.32.1h
Sol medium
120.0$17.31.5h
Sol low
90.0$9.40.9h
Terra max
90.0$37.34.3h
Terra high
105.0$10.11.4h
Luna max
120.0$14.84.4h
Luna high
105.0$6.31.9h
150 150 120 120 90 90 60 60 90-110 band 7.10_n 7.11_pm 7.12_am 7.12_pm 7.13_am 7.13_pm 7.14_am 7.14_pm 7.15_am 7.15_pm 7.16_am
Daily metrics
MetricSol maxSol xhighSol highSol mediumSol lowTerra maxTerra highLuna maxLuna high
Passes7/107/107/108/106/106/107/108/107/10
IQ105.0105.0105.0120.090.090.0105.0120.0105.0
Agent steps69.152.238.034.521.683.236.976.147.1
Cost$60.2$35.6$23.3$17.3$9.4$37.3$10.1$14.8$6.3
Cache hit rate97.0%96.6%95.5%94.5%90.9%97.7%95.1%96.5%96.3%
Time4.9h2.9h2.1h1.5h0.9h4.3h1.4h4.4h1.9h
Total tokens70.5M39.3M22.6M16.4M7.3M90.5M19.6M82.7M34.7M
Fixed task set

Fixed DeepSWE task set

The daily IQ point is scored on this fixed mixed-language DeepSWE subset. Each row shows today's result for the tracked models.

10 tasks
  1. 01
    Add JSONPath query APIs to orderedmap and Starlark modules ytt-jsonpath-query-api
    GO history pass 79% (284/359)
    Sol max pass Sol xhigh pass Sol high pass Sol medium pass Sol low pass Terra max fail Terra high pass Luna max pass Luna high pass
  2. 02
    Harden module loading, cache introspection, and script flags abs-module-cache-flags
    GO history pass 79% (284/359)
    Sol max fail Sol xhigh fail Sol high fail Sol medium fail Sol low pass Terra max pass Terra high pass Luna max pass Luna high pass
  3. 03
    Add multipart response parsing to HTTPX httpx-multipart-response-parsing
    PYTHON history pass 94% (339/359)
    Sol max pass Sol xhigh pass Sol high pass Sol medium pass Sol low pass Terra max pass Terra high pass Luna max pass Luna high pass
  4. 04
    Add incremental cache controls to Bandit bandit-incremental-cache-control
    PYTHON history pass 41% (147/359)
    Sol max pass Sol xhigh pass Sol high pass Sol medium pass Sol low fail Terra max pass Terra high pass Luna max pass Luna high pass
  5. 05
    Add session bundle recording and replay to IPython ipython-session-bundle-replay
    PYTHON history pass 39% (141/359)
    Sol max pass Sol xhigh pass Sol high pass Sol medium pass Sol low fail Terra max pass Terra high fail Luna max pass Luna high pass
  6. 06
    Add a per-origin circuit breaker to ofetch ofetch-per-origin-circuit-breaker
    TYPESCRIPT history pass 94% (338/359)
    Sol max pass Sol xhigh pass Sol high pass Sol medium pass Sol low pass Terra max pass Terra high pass Luna max pass Luna high pass
  7. 07
    Add link format conversion between wiki and markdown syntax obsidian-linter-link-format-conversion
    TYPESCRIPT history pass 17% (60/359)
    Sol max fail Sol xhigh fail Sol high fail Sol medium fail Sol low fail Terra max fail Terra high fail Luna max pass Luna high fail
  8. 08
    Add shorthand expansion and compression to the lexer csstree-shorthand-expansion-compression
    JAVASCRIPT history pass 50% (181/359)
    Sol max pass Sol xhigh pass Sol high pass Sol medium pass Sol low fail Terra max fail Terra high pass Luna max fail Luna high fail
  9. 09
    Add deterministic multi-key sorting to fd fd-deterministic-multi-key-sorting
    RUST history pass 42% (150/359)
    Sol max pass Sol xhigh pass Sol high pass Sol medium pass Sol low pass Terra max pass Terra high pass Luna max pass Luna high pass
  10. 10
    Preserve structure needed by stylesheet selectors oxvg-structural-selector-preservation
    RUST history pass 60% (216/358)
    Sol max fail Sol xhigh fail Sol high fail Sol medium pass Sol low pass Terra max fail Terra high fail Luna max fail Luna high fail
Previous 5.5 / 5.4 history

Archived previous-generation curves and daily comparison table. This section is no longer updated.

150 150 120 120 90 90 60 60 90-110 band 5.30 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 6.10 6.11 6.12 6.13 6.14 6.15_am 6.15_pm 6.16_am 6.16_pm 6.17_am 6.17_pm 6.18 6.19 6.20 6.21 6.22_am 6.22_pm 6.23 6.24_am 6.24_pm 6.25 6.25_pm 6.26 6.26_pm 6.27_am 6.27_pm 6.28_am 6.28_pm 6.29_am 6.29_pm 6.30_am 6.30_pm 7.1_am 7.1_pm 7.2 7.2_pm 7.2_pm_2 7.2_n 7.3_am 7.3_pm 7.4_am 7.4_pm 7.5_am 7.5_pm 7.6_am 7.6_pm 7.7_am 7.7_pm 7.8_am 7.8_pm 7.9_am 7.9_pm
Metric5.5-xhigh5.5-high5.5-medium5.5-low5.4-xhigh5.4-high
Average Passes6.0/106.4/105.8/105.2/105.2/105.4/10
Average IQ90.596.186.878.078.381.4
Average Agent steps41.733.931.227.144.737.5
Average Cost$35.7$23.4$18.7$13.9$19.3$13.8
Average Cache hit rate94.8%93.8%93.3%92.9%95.7%95.0%
Average Time2.4h1.6h1.3h1.0h2.8h2.0h
Average Total tokens36.0M22.9M18.2M13.4M38.9M27.2M

Score

Reference: 9-10 clearly strong, 7-8 normally usable, 5-6 barely usable, 3-4 poor, 1-2 nearly unusable. Consider accuracy, rework, speed, stability, and quota/service impact.

Last 24h rating
ModelIn/C/Out
ultra
max
xhigh
high
medium
low
Sol$5 · $0.50 · $305/0.5/30
GPT-5.6 Sol ultra
N/A 0 ratings
GPT-5.6 Sol max
N/A 0 ratings
GPT-5.6 Sol xhigh
N/A 0 ratings
GPT-5.6 Sol high
N/A 0 ratings
GPT-5.6 Sol medium
N/A 0 ratings
GPT-5.6 Sol low
N/A 0 ratings
Terra$2.50 · $0.25 · $152.5/0.25/15
GPT-5.6 Terra ultra
N/A 0 ratings
GPT-5.6 Terra max
N/A 0 ratings
GPT-5.6 Terra xhigh
N/A 0 ratings
GPT-5.6 Terra high
N/A 0 ratings
GPT-5.6 Terra medium
N/A 0 ratings
GPT-5.6 Terra low
N/A 0 ratings
Luna$1 · $0.10 · $61/0.1/6
Not supported
GPT-5.6 Luna max
N/A 0 ratings
GPT-5.6 Luna xhigh
N/A 0 ratings
GPT-5.6 Luna high
N/A 0 ratings
GPT-5.6 Luna medium
N/A 0 ratings
GPT-5.6 Luna low
N/A 0 ratings
Tibo Radar
Tibo's X avatar
--:-- -- Calculating
San Francisco Bay Area / PT
Coarse public time-zone estimate The recent public posts do not self-disclose a country or region, so the default time zone is retained. Source ↗