subsdaddy
Subscription plans compared on $/B, tok/s and TTFT p50.
as of 2026-10-28T01:50:09.614Z · usage window 31 days from 2026-10-07T18:36:26Z · OpenRouter stats fetched 2026-10-28T01:50:09.800Z
ours, from usage logs
| Plan | Fee/mo | Allowance | Capacity | $/B | Threads est | tok/s | TTFT p50 |
|---|---|---|---|---|---|---|---|
| OpenCode Go | $10.00/mo | $80.00/30d(p) | 2.07B/30dcache 87.9% | $4.83 | 0.9 | ~180 | ~2,500 ms |
| Z.ai Coding Plan - Max (Legacy v2) | $56.00/mo | $106.66/7d(d) | 2.06B/7dcache 89.8%low confidencestale reading | $6.24 | 4.0 | ~67 | ~6,900 ms |
| QwenCloud - Token Plan Individual | $10.00/mo | $20.26/30d(c) | 0.35B/30dcache 92.9% | $28.51 | 0.2 | ~140 | ~3,900 ms |
| CommandCode - GOAT | $10.00/mo | N/A | N/A | N/A | N/A | ~200 | ~3,700 ms |
Measured $/B is per plan and reflects that plans own cache share; use Shape to compare on equal terms.
Allowance source: p published (none) · r vendor reading (none) · d derived from quota windows (Z.ai Coding Plan - Max (Legacy v2) $106.66) · c credit value (QwenCloud - Token Plan Individual $0.000794 per credit) · t third-party measured weekly value × 30/7 · j third-party projection, not a measurement
Bring your own shape: a local collector is planned; the site will price every plan against your own cache share.
Estimates assume the whole allowance runs one model. Details, sources and allowances: /methodology