You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit 4f32280
Browse filesBrowse the repository at this point in the historyBrowse files
Profile ids are immutable after creation. To use a different id, create a new profile and remove the old one after updating callers.
@@ -28,6 +28,14 @@ The dashboard sends the same profile object used by `config.routingProfiles` to
28
28
29
29
A successful save persists the profile through the normal config writer, reconciles live state, and refreshes the model catalog. Validation failures leave the previous configuration unchanged and are shown in the editor.
30
30
31
+
When `limits.maxEstimatedCostUsd` is configured, `limits.onUnknownCost` defaults to `"allow"`: an unknown cost estimate does not get a
32
+
cap-specific exclusion, and dry-run / live route-decision traces stamp
33
+
`cost.capOutcome: "unknown-allowed"` so operators can tell the cap was not proven. Set `"exclude"`
34
+
when the ceiling must fail closed (`cost-limit-unknown`, with
35
+
`cost.capOutcome: "unknown-excluded"`). Configuring `onUnknownCost` alone is inert and does not emit a cap outcome. This is separate from
36
+
`unknownEvidence.cost`, which can still exclude or penalize unknown prices independently of the
37
+
cap outcome.
38
+
31
39
## Dry-run a saved profile
32
40
33
41
Select a saved profile and use **Dry-run evaluation** to add request evidence such as context-window size, tool use, image input, or structured output. Dry-run evaluates eligibility and scoring but never sends an upstream model request.
Copy file name to clipboardExpand all lines: docs-site/src/content/docs/reference/configuration/routing.md
+8-4Lines changed: 8 additions & 4 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -114,7 +114,7 @@ namespace, or reserved bare native families (`gpt-*`, `o1-*`, `o3-*`, `o4-*`, `c
114
114
|`alias?`|`string`| — | Optional public model id in place of `policy/<id>`. |
115
115
|`require?`| object |`{}`| Hard capability requirements evaluated before scoring (see below). |
116
116
|`optimize?`| object | latency 0.55, health 0.25, cost 0.10, quota 0.10 | Scoring weights, normalized deterministically. `health`, `quota`, and `cost` have score dimensions; the configured-priority share is `1 - health - quota - cost` (default 0.55), and `latency` folds into that priority share rather than scoring independently. |
117
-
|`limits?`| object | — | Hard limits, e.g.`maxEstimatedCostUsd`(enforced by the dry-run evaluator when candidate cost evidence is known). |
117
+
|`limits?`| object | — | Hard limits.`maxEstimatedCostUsd`excludes a candidate when its estimated cost is known and above the cap. When that cap is set, `onUnknownCost` (`"allow"` default, or `"exclude"`) controls unknown estimates: allow prevents a cap-specific exclusion and records `cost.capOutcome: "unknown-allowed"`; exclude emits `cost-limit-unknown` and `capOutcome: "unknown-excluded"`. `onUnknownCost` alone (no cap) is inert. Separate from `unknownEvidence.cost`, which can still exclude or penalize unknown prices via `unknown-price` / scoring. |
118
118
|`unknownEvidence?`| object | capability `exclude`, health/quota/cost `penalize`| How unknown evidence is treated per dimension: `allow`, `penalize`, or `exclude`. Unknown never becomes zero. |
0 commit comments