amadan
🤖 spec: price the cheap tier, closing §8.6's last owed figure
deepseek/deepseek-v4-flash-0731 is $0.25/$0.30 per M, so ~$0.0014 a query against §8.6's 5k-in/500-out shape, and ~$3/month across 2,200 queries. With v4-pro-0813 at ~$26/month all-frontier, both ends of the estimate are now derived from the ids §8.2 actually pins rather than from a comparable. One correction to how this section reads its own history. It said the TensorX switch dropped the ceilings while leaving the estimates inside their old ranges. That was true of the glm/kimi pair; it is not true now. The frontier figure at $0.012 falls below the $0.02-0.10 range this section carried rather than sitting in it, so the ceiling moved by about 4x. The cause is §8.2 pinning dated ids from one family, not the provider switch — worth separating, because attributing a 4x to the wrong decision is how the next provider argument gets made on a false premise. §13 loses the unpriced item. What remains needs someone other than me: the embeddings batch size and a re-probe of the 403 need a key, id immutability needs an email to the vendor, and the subprocessor question needs the DPA.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
1 file changed,
+13
−14
docs/superpowers/specs/2026-09-03-memoria-design.md+13 −14
diff --git a/docs/superpowers/specs/2026-09-03-memoria-design.md b/docs/superpowers/specs/2026-09-03-memoria-design.md| index aff242f..8adff7e 100644 |
| --- a/docs/superpowers/specs/2026-09-03-memoria-design.md |
| +++ b/docs/superpowers/specs/2026-09-03-memoria-design.md |
| @@ -1442,18 +1442,19 @@ from §7.2's measurement, and if that measurement moves so does this. |
| **Query:** retrieval is free. The generative layer sends ~8 passages plus |
| instructions — ~5k tokens in, ~500 out. Against the ids §8.2 pins: |
| -`deepseek/deepseek-v4-pro-0813` at $2.00/$4.00 per M is **~$0.012 a |
| -query**, so 5 people × 20 queries/day × 22 days = 2,200 queries is |
| -**~$26/month with everything on the frontier tier**. **`deepseek/deepseek-v4-flash-0731` |
| -is unpriced here** — it was pinned after the price list was read, and the |
| -figure is owed (§13). For scale only, and not as an estimate: the |
| -comparably-named `z-ai/glm-5.3-flash` is $0.20/$0.50, which would put the |
| -all-cheap month near $3. Read the rest carefully: |
| -both figures sit *inside* the $0.001–0.003 and $0.02–0.10 per-query ranges |
| -this section already carried, near the bottom. **The ceilings fell; the |
| -estimates did not move.** Nothing here is a reason to route more queries to |
| -the frontier tier — §8.2's three model-free classes are still the design, |
| -and §8.6's closing sentence still holds. |
| +`deepseek/deepseek-v4-flash-0731` at $0.25/$0.30 per M is **~$0.0014 a |
| +query**, and `deepseek/deepseek-v4-pro-0813` at $2.00/$4.00 is |
| +**~$0.012**. Five people × 20 queries/day × 22 days = 2,200 queries: |
| +**~$3/month if most route cheap, ~$26 if most route frontier.** Read that |
| +carefully: |
| +the cheap figure sits *inside* the $0.001–0.003 per-query range this |
| +section already carried; the frontier one, at $0.012, finally falls below |
| +its $0.02–0.10 range rather than merely sitting in it. **So the ceiling |
| +fell by about 4× and the floor did not move.** That is a consequence of |
| +§8.2 pinning dated ids from one family, not of the provider switch, and it |
| +is still not a reason to route more queries to the frontier tier — §8.2's |
| +three model-free classes remain the design, and this section's closing |
| +sentence still holds. |
| Derived-artefact regeneration (§5.2) adds a synthesis-class call per |
| invalidated artefact and is unbudgeted until the feature exists. |
| @@ -1843,8 +1844,6 @@ rows. |
| whether a dated id is itself immutable at TensorX (ask them), and what |
| happens to the **embedding** model, which has no dated form at all — gate |
| 24's canary is the only detector for that one. |
| -- **`deepseek/deepseek-v4-flash-0731` is unpriced**, and §8.6's cheap-tier |
| - month cannot be stated until it is. One lookup. |
| - **Both chat tiers now come from one lab as well as one host.** §8.2 took |
| dated ids over family preference, correctly under its own rule, but a |
| DeepSeek withdrawal would move both tiers at once where the previous pair |