* feat: support tencent tokenhub api key via openai-compatible protocol
* fix: use tokenhub base url for tencent api key channels
* test: cover tencent key-format dispatch and add TokenHub key prompt locales
* fix: prevent model create from wiping existing pricing for same name
* fix(models): keep create from wiping pricing for any existing name
Prefilling from the drawer's open-time model name only covered the
missing-models entry point. Submit deletes the pricing entries for the
name in the form, which is editable and starts empty from the toolbar
"Create model" button, so both a hand-typed existing name and a renamed
prefill still dropped the configured pricing.
Track the name whose pricing was actually read into the form and scope
the delete-then-readd to it, or to a name the user explicitly priced.
Editing still clears pricing by emptying the fields, a prefilled create
does too, and a name the form never loaded is left alone -- which also
stops an edit that renames onto an existing name from destroying that
name's pricing.
Read the prefill through one shared readPricingConfig instead of
duplicating the seven-map parse in both branches, and open the advanced
section for ratios that are configured as 0 rather than only truthy ones.
controller/task_video.go and service/pre_consume_quota.go were already
removed in ba25ba88f and 116004fd4 (logic lives in service/task_polling.go
and BillingSession now), then brought back as stale copies by a42b39760.
Both have zero callers.
controller/swag_video.go is a leftover swag stub: the swaggo pipeline is
gone and docs/openapi/relay.json already documents these routes.
The ratio column normalized input to number on every keystroke, so
typing 0.05 got stuck at 0.0 which Number() collapses to 0. Keep the
value as a string draft like the top-up ratio column and only convert
to number when serializing.
Fixes#5986
* feat: add graceful shutdown with configurable timeout
* fix: flush quota dashboard cache on graceful shutdown
Persist the in-memory CacheQuotaData aggregation to the quota_data table
before process exit, so a restart no longer drops up to one DataExportInterval
window of dashboard data (issue #5679).
AnimatedOutlet keyed its motion container by the full pathname, so
navigating between params of the same route (e.g. the dashboard tabs
under /dashboard/$section) changed the key and remounted the route
component, discarding its in-memory state. The most visible symptom:
the dashboard time range resets to the default every time you switch
sub-tabs (Model Analytics / Flow / User Analytics).
Key the transition by the matched route id instead, so param-only
navigation re-renders in place while real page-to-page navigation
still plays the enter animation. This also preserves state across the
other $section tab pages (usage logs, models, system settings).
Async task usage logs (LogQuotaData node dimension) were recorded
under whichever node happened to poll the task to completion, not the
node that submitted it. For token/adaptor-billed video tasks the
pre-deduction is often 0, so the entire quota landed on the last
polling node.
Snapshot common.NodeName into TaskPrivateData at submit time and use
it when writing the settlement consume log; fall back to the current
node when empty so existing tasks stay compatible.
Switching models in the pricing editor kept the previous model's tiers and prices in the expression panel: TieredPricingEditor seeds its internal visual/raw state only on mount, and the initRef guard never re-ran on prop changes, so only the model name updated.
Bump a reload token in the same effect that seeds billingExpr and use it as the editor's key, so a freshly loaded model remounts the editor and re-parses its expression. The token changes in lockstep with billingExpr, and user edits (which only touch state) do not trigger it.
Closes#5750
The profile dropdown rendered the wallet item unconditionally, so it
still showed after an admin disabled the personal/topup (wallet) sidebar
module. Reuse the sidebar module visibility check so the dropdown honours
the same toggle as the sidebar.
Fixes#5696
Fall back to the OS hostname when NODE_NAME is not configured, so node
identity in audit/usage logs is populated automatically. In container and
Kubernetes deployments the hostname equals the container ID or Pod name,
which stays unique under autoscaling without any manual per-instance config.
feitianbubu