omk-ai 1.5.1 → 1.5.3
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +12 -2
- package/dist/cli.js +0 -0
- package/dist/models.generated.d.ts.map +1 -1
- package/dist/models.generated.js +2 -2
- package/dist/models.generated.js.map +1 -1
- package/package.json +1 -1
package/CHANGELOG.md
CHANGED
|
@@ -1,13 +1,23 @@
|
|
|
1
1
|
# Changelog
|
|
2
2
|
|
|
3
|
+
## [1.5.3] - 2026-10-19
|
|
4
|
+
|
|
5
|
+
### Changed
|
|
6
|
+
|
|
7
|
+
- WorkBuddy's DeepSeek V4.1 Flash lane is labeled 1017, uses the current `@tencent-ai/codebuddy-code@2.169.0` product cap of 128k output tokens, and keeps `low`, `high`, and `max` effort selection. A live probe on 2026-10-19 accepted `low`, `high`, `max`, and `xhigh`; the dated `*-1017` and `deepseek-v4-flash` ids answered `400 11102`, so the undated product id remains the declared lane.
|
|
8
|
+
|
|
9
|
+
## [1.5.2] - 2026-10-18
|
|
10
|
+
|
|
11
|
+
### Changed
|
|
12
|
+
|
|
13
|
+
- Lockstep version alignment with OMK 1.5.2 to restore publication of all seven workspace packages after the 1.5.1 registry staging conflict. Provider recovery behavior is unchanged from 1.5.1.
|
|
14
|
+
|
|
3
15
|
## [1.5.1] - 2026-10-18
|
|
4
16
|
|
|
5
17
|
### Fixed
|
|
6
18
|
|
|
7
19
|
- Provider retry classification now treats Cloudflare `520` through `527` as transient and recognizes Explabs/OpenRouter quota prefixes (`free_limit_reached`, `insufficient_credits`, `model_requires_purchase`, `promo_byok_only`, `no available platform credits`). The observed `520 status code (no body)` and free-limit bodies no longer end a summarization or agent turn without a retry decision.
|
|
8
20
|
|
|
9
|
-
## [Unreleased]
|
|
10
|
-
|
|
11
21
|
## [1.5.0] - 2026-10-18
|
|
12
22
|
|
|
13
23
|
- Experiential Labs (`experientiallabs`) joins the catalog as an OpenAI-compatible gateway at `https://api.experientiallabs.ai/v1`, authenticated by `EXPLABS_API_KEY` or `EXPERIENTIAL_API_KEY`. The 235 callable, routable, tool-capable text models are generated from the public `/api/models` catalog plus the authenticated `/v1/models` surface (checked 2026-10-18). Reasoning uses the `/v1/models` contract, not the upstream route advertisement: 150 models expose a `thinkingLevelMap`, `none` maps to off, and unsupported tiers are hidden. This keeps GLM-5.2 provider-default because its gateway surface rejects `reasoning_effort`, while GPT-6 Sol and Luna expose the gateway's none/low/medium/high/xhigh/max ladder. Media-only, embedding-only, non-tool and non-callable rows stay out because the text catalog cannot route them.
|
package/dist/cli.js
CHANGED
|
File without changes
|