codex · gpt-5.6-sol · low
89.5overallpartial
Overall rank
#1 of 2
Cost efficiency rank
#2 of 2
Wall time rank
#2 of 2
Telemetry
partial
codex · gpt-5.6-sol · low leads this release at 89.5 overall. At 18.45 API-equivalent it costs 14.1x claude · haiku · low, the cheapest scored agent this release.
Tokens by class
Segment width is dollar share, not token share — where the API-equivalent cost actually went.
Input 3.9M $15.53Cache read 3.6M $1.44Output 74.1K $1.48
Tokens in
3.9M
Tokens out
74.1K
Turns
7
Wall time
28m 31s
API-equiv. cost
$18.45
Score / $
4.8
Category scores
UX/UI
100.0 (1 tasks)
Frontend
68.3 (1 tasks)
Backend
— (1 tasks)
Planning Audit
79.0 (1 tasks)
Bug Fix
100.0 (1 tasks)
Taste
100.0 (1 tasks)
Signature
92.5 (1 tasks)
Per task
| Task | Gate pass | Objective | Subjective | Score | Runs |
|---|---|---|---|---|---|
| Issue board filters | 100% | 55.6 (55.6–55.6, n=1) | 87.5 (87.5–87.5, n=1) | 68.3 (68.3–68.3, n=1) | |
| shipshit.dev landing hero | 100% | — | 92.5 (92.5–92.5, n=1) | 92.5 (92.5–92.5, n=1) | |
| Codebase audit, planted defects | 100% | 65.0 (65.0–65.0, n=1) | 100.0 (100.0–100.0, n=1) | 79.0 (79.0–79.0, n=1) | |
| Inventory service | 0% | — | — | — | |
| Shipcut pricing page | 100% | — | 100.0 (100.0–100.0, n=1) | 100.0 (100.0–100.0, n=1) | |
| Landing page, five themes | 100% | — | 100.0 (100.0–100.0, n=1) | 100.0 (100.0–100.0, n=1) | |
| Fix date range overlap bug | 100% | 100.0 (100.0–100.0, n=1) | — | 100.0 (100.0–100.0, n=1) |