Skip to content

Get up and running with Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.

server: proxy cloud usage and balance APIs (#18829)

Devon Rifkinf864601538bd
diff --git a/docs/api/balance.mdx b/docs/api/balance.mdx
new file mode 100644
index 000000000..2d6c40faa
--- /dev/null
+++ b/docs/api/balance.mdx
@@ -0,0 +1,78 @@
+---
+title: Balance
+sidebarTitle: /api/balance
+openapi: /openapi.yaml GET /api/balance
+---
+
+<RequestExample>
+
+```shell Request
+curl https://ollama.com/api/balance \
+ -H "Authorization: Bearer $OLLAMA_API_KEY"
+```
+
+</RequestExample>
+
+<ResponseExample>
+
+```json Response
+{
+ "included": {
+ "balance_usd": 72.5,
+ "allowance_usd": 100,
+ "period": {
+ "from": "2026-09-15T09:30:00Z",
+ "until": "2026-10-15T09:30:00Z"
+ }
+ },
+ "purchased": {
+ "balance_usd": 25
+ }
+}
+```
+
+</ResponseExample>
+
+This endpoint takes no query parameters. For usage history, see [Cloud usage](/api/cloud-usage).
+
+## Balances
+
+The included period follows your plan's monthly reset schedule, including for annual subscriptions. Period timestamps are in UTC, with `from` inclusive and `until` exclusive.
+
+The purchased balance includes only unexpired credits.
+
+<Accordion title="Legacy plans">
+
+Plans with session and weekly limits return `session` and `weekly` in `included`:
+
+```json
+{
+ "included": {
+ "session": {
+ "remaining_percent": 75,
+ "resets_at": "2026-10-01T07:00:00Z"
+ },
+ "weekly": {
+ "remaining_percent": 40,
+ "resets_at": "2026-10-05T00:00:00Z"
+ }
+ },
+ "purchased": {
+ "balance_usd": 25
+ }
+}
+```
+
+`remaining_percent` ranges from `0` to `100`; `75` means 75% remains. `resets_at` is the next reset time in UTC.
+
+</Accordion>
+
+## Rate limits
+
+This endpoint allows 10 requests per minute per user, shared across API keys and devices. We recommend polling once per minute. A `429` response includes a `Retry-After` header with the number of seconds to wait before retrying.
+
+## Teams
+
+All team members share the same included and purchased balances.
+
+## Reference
diff --git a/docs/api/cloud-usage.mdx b/docs/api/cloud-usage.mdx
new file mode 100644
index 000000000..9fd3cf9b0
--- /dev/null
+++ b/docs/api/cloud-usage.mdx
@@ -0,0 +1,120 @@
+---
+title: Cloud usage
+sidebarTitle: /api/usage
+openapi: /openapi.yaml GET /api/usage
+---
+
+<RequestExample>
+
+```shell Request
+curl 'https://ollama.com/api/usage?range=24h' \
+ -H "Authorization: Bearer $OLLAMA_API_KEY"
+```
+
+</RequestExample>
+
+<ResponseExample>
+
+```json Response (22 empty buckets omitted)
+{
+ "range": "24h",
+ "scope": "self",
+ "granularity": "hour",
+ "from": "2026-09-30T02:00:00Z",
+ "until": "2026-10-01T02:30:00Z",
+ "totals": {
+ "request_count": 15,
+ "usage_usd": 0.01718,
+ "input_tokens": 106000,
+ "cached_input_tokens": 46000,
+ "output_tokens": 13600
+ },
+ "buckets": [
+ {
+ "from": "2026-09-30T12:00:00Z",
+ "until": "2026-09-30T13:00:00Z",
+ "request_count": 4,
+ "usage_usd": 0.0052,
+ "input_tokens": 24000,
+ "cached_input_tokens": 0,
+ "output_tokens": 3200
+ },
+ {
+ "from": "2026-10-01T00:00:00Z",
+ "until": "2026-10-01T01:00:00Z",
+ "request_count": 8,
+ "usage_usd": 0.0088,
+ "input_tokens": 64000,
+ "cached_input_tokens": 40000,
+ "output_tokens": 8000
+ },
+ {
+ "from": "2026-10-01T02:00:00Z",
+ "until": "2026-10-01T02:30:00Z",
+ "partial": true,
+ "request_count": 3,
+ "usage_usd": 0.00318,
+ "input_tokens": 18000,
+ "cached_input_tokens": 6000,
+ "output_tokens": 2400
+ }
+ ]
+}
+```
+
+</ResponseExample>
+
+New usage may take time to appear.
+
+For your remaining included and purchased credits, use [Balance](/api/balance).
+
+Unsupported or repeated query parameters return `400`. Errors follow Ollama's [JSON error format](/api/errors#error-messages).
+
+## Time ranges
+
+| Range | Starts at | Bucket size | Buckets |
+| ----- | ------------------------------------------------- | ----------- | ------- |
+| `24h` | 24 hours before the start of the current UTC hour | Hour | 25 |
+| `7d` | Seven days ago at 00:00 UTC | Day | 8 |
+| `30d` | Thirty days ago at 00:00 UTC | Day | 31 |
+
+Every range ends at the time of the request. For example, at 02:30 UTC on October 1, `24h` starts at 02:00 UTC on September 30. It returns 24 complete hourly buckets followed by the current 02:00–02:30 bucket. The `7d` and `30d` ranges return seven and thirty complete days, followed by today so far.
+
+Buckets appear in chronological order, including zeroes for hours or days with no recorded requests. The final bucket covers the current hour or day and has `partial: true`.
+
+All timestamps are in UTC. Each `from` is inclusive and each `until` is exclusive.
+
+## Usage totals and buckets
+
+`totals` summarizes usage across the entire range. `buckets` breaks that usage into hours or days.
+
+`granularity` is set automatically: `hour` for `24h` and `day` for `7d` and `30d`. The top-level `from` and `until` describe the entire range; each bucket has its own boundaries.
+
+`partial` indicates whether the hour or day is still in progress, rather than whether all usage has been recorded.
+
+<Accordion title="Legacy plans">
+
+For legacy plans, only request counts are available. If a bucket or total includes legacy requests, cost and token counts are omitted.
+
+</Accordion>
+
+## Rate limits
+
+This endpoint allows 10 requests per minute per user, shared across API keys and devices. We recommend polling once per minute. A `429` response includes a `Retry-After` header with the number of seconds to wait before retrying.
+
+## Team usage
+
+By default, everyone sees their own usage (`scope=self`).
+
+Team admins can use `scope=team` to see usage for the whole team:
+
+```shell
+curl 'https://ollama.com/api/usage?range=7d&scope=team' \
+ -H "Authorization: Bearer $OLLAMA_API_KEY"
+```
+
+## Coming soon
+
+We plan to add custom time ranges, usage breakdowns by model and API key, and breakdowns by team member for team admins.
+
+## Reference
diff --git a/docs/docs.json b/docs/docs.json
index 44a655bcd..ed96b7167 100644
--- a/docs/docs.json
+++ b/docs/docs.json
@@ -259,7 +259,9 @@
"POST /api/pull",
"POST /api/push",
"DELETE /api/delete",
- "GET /api/version"
+ "GET /api/version",
+ "/api/cloud-usage",
+ "/api/balance"
]
}
]
diff --git a/docs/openapi.yaml b/docs/openapi.yaml
index c66ffaa40..675c8bb59 100644
--- a/docs/openapi.yaml
+++ b/docs/openapi.yaml
@@ -1017,6 +1017,135 @@ components:
type: integer
minimum: 0
description: Tokens generated internally for scoring, including prefix preparation and retries. May exceed the question count; not the length of the JSON response.
+ UsageMetrics:
+ type: object
+ required: [request_count]
+ properties:
+ request_count:
+ type: integer
+ minimum: 0
+ description: Number of recorded requests.
+ usage_usd:
+ type: number
+ description: Total value of the requests in USD, including usage covered by your plan and usage paid from purchased credits.
+ input_tokens:
+ type: integer
+ description: Recorded input tokens, including cached input tokens.
+ cached_input_tokens:
+ type: integer
+ description: Recorded input tokens read from cache, already included in input_tokens.
+ output_tokens:
+ type: integer
+ description: Recorded output tokens.
+ UsageBucket:
+ allOf:
+ - $ref: "#/components/schemas/UsageMetrics"
+ - type: object
+ required: [from, until]
+ properties:
+ from:
+ type: string
+ format: date-time
+ description: Start of the bucket in UTC, inclusive.
+ until:
+ type: string
+ format: date-time
+ description: End of the bucket in UTC, exclusive.
+ partial:
+ type: boolean
+ description: Present and true for the current hour or day, which is still in progress.
+ UsageResponse:
+ type: object
+ required: [range, scope, granularity, from, until, totals, buckets]
+ properties:
+ range:
+ type: string
+ enum: [24h, 7d, 30d]
+ description: Selected time range.
+ scope:
+ type: string
+ enum: [self, team]
+ description: Selected usage scope.
+ granularity:
+ type: string
+ enum: [hour, day]
+ description: "Set automatically: hour for 24h, day for 7d and 30d."
+ from:
+ type: string
+ format: date-time
+ description: Start of the range in UTC, inclusive.
+ until:
+ type: string
+ format: date-time
+ description: End of the range in UTC, exclusive.
+ totals:
+ $ref: "#/components/schemas/UsageMetrics"
+ description: Usage across the entire range.
+ buckets:
+ type: array
+ description: Usage in chronological order, including buckets with no recorded requests.
+ items:
+ $ref: "#/components/schemas/UsageBucket"
+ IncludedBalance:
+ title: Included credits
+ type: object
+ required: [balance_usd, allowance_usd, period]
+ properties:
+ balance_usd:
+ type: number
+ description: Remaining included credits in USD.
+ allowance_usd:
+ type: number
+ description: Included credits available for the full period in USD.
+ period:
+ type: object
+ required: [from, until]
+ properties:
+ from:
+ type: string
+ format: date-time
+ description: Start of the included period in UTC, inclusive.
+ until:
+ type: string
+ format: date-time
+ description: End of the included period in UTC, exclusive.
+ LegacyBalanceLimit:
+ type: object
+ required: [remaining_percent, resets_at]
+ properties:
+ remaining_percent:
+ type: number
+ minimum: 0
+ maximum: 100
+ description: Percentage of the limit remaining.
+ resets_at:
+ type: string
+ format: date-time
+ description: Next reset time in UTC.
+ LegacyIncludedBalance:
+ title: Legacy plan limits
+ type: object
+ required: [session, weekly]
+ properties:
+ session:
+ $ref: "#/components/schemas/LegacyBalanceLimit"
+ weekly:
+ $ref: "#/components/schemas/LegacyBalanceLimit"
+ BalanceResponse:
+ type: object
+ required: [included, purchased]
+ properties:
+ included:
+ oneOf:
+ - $ref: "#/components/schemas/IncludedBalance"
+ - $ref: "#/components/schemas/LegacyIncludedBalance"
+ purchased:
+ type: object
+ required: [balance_usd]
+ properties:
+ balance_usd:
+ type: number
+ description: Remaining unexpired purchased credits in USD.
ErrorResponse:
type: object
properties:
@@ -1034,7 +1163,7 @@ paths:
href: /api/systemone
metadata:
title: System One
- sidebarTitle: System One
+ sidebarTitle: /v1/systemone
content: |
<Note>
Currently available locally in Ollama.
@@ -1133,6 +1262,8 @@ paths:
description: Generates a response for the provided prompt
operationId: generate
x-mint:
+ metadata:
+ sidebarTitle: /api/generate
href: /api/generate
x-codeSamples:
- lang: bash
@@ -1254,6 +1385,8 @@ paths:
description: Generate the next chat message in a conversation between a user and an assistant.
operationId: chat
x-mint:
+ metadata:
+ sidebarTitle: /api/chat
href: /api/chat
x-codeSamples:
- lang: bash
@@ -1412,6 +1545,8 @@ paths:
description: Creates vector embeddings representing the input text
operationId: embed
x-mint:
+ metadata:
+ sidebarTitle: /api/embed
href: /api/embed
x-codeSamples:
- lang: bash
@@ -1487,6 +1622,8 @@ paths:
description: Fetch a list of models and their details
operationId: list
x-mint:
+ metadata:
+ sidebarTitle: /api/tags
href: /api/tags
x-codeSamples:
- lang: bash
@@ -1520,6 +1657,8 @@ paths:
description: Retrieve a list of models that are currently running
operationId: ps
x-mint:
+ metadata:
+ sidebarTitle: /api/ps
href: /api/ps
x-codeSamples:
- lang: bash
@@ -1554,6 +1693,9 @@ paths:
post:
summary: Show model details
operationId: show
+ x-mint:
+ metadata:
+ sidebarTitle: /api/show
x-codeSamples:
- lang: bash
label: Default
@@ -1678,6 +1820,8 @@ paths:
summary: Create a model
operationId: create
x-mint:
+ metadata:
+ sidebarTitle: /api/create
href: /api/create
x-codeSamples:
- lang: bash
@@ -1790,6 +1934,8 @@ paths:
summary: Copy a model
operationId: copy
x-mint:
+ metadata:
+ sidebarTitle: /api/copy
href: /api/copy
x-codeSamples:
- lang: bash
@@ -1834,6 +1980,8 @@ paths:
summary: Pull a model
operationId: pull
x-mint:
+ metadata:
+ sidebarTitle: /api/pull
href: /api/pull
x-codeSamples:
- lang: bash
@@ -1876,6 +2024,8 @@ paths:
summary: Push a model
operationId: push
x-mint:
+ metadata:
+ sidebarTitle: /api/push
href: /api/push
x-codeSamples:
- lang: bash
@@ -1918,6 +2068,8 @@ paths:
summary: Delete a model
operationId: delete
x-mint:
+ metadata:
+ sidebarTitle: /api/delete
href: /api/delete
x-codeSamples:
- lang: bash
@@ -1942,6 +2094,9 @@ paths:
summary: Get version
description: Retrieve the version of the Ollama
operationId: version
+ x-mint:
+ metadata:
+ sidebarTitle: /api/version
x-codeSamples:
- lang: bash
label: Default
@@ -1956,3 +2111,124 @@ paths:
$ref: "#/components/schemas/VersionResponse"
example:
version: "0.12.6"
+ /api/usage:
+ get:
+ summary: Cloud usage
+ description: Usage statistics for cloud inference, web search, and web fetch.
+ operationId: usage
+ servers:
+ - url: https://ollama.com
+ description: Ollama Cloud
+ security:
+ - bearerAuth: []
+ parameters:
+ - name: range
+ in: query
+ description: Completed UTC hours or days to include, plus the current partial hour or day.
+ schema:
+ type: string
+ enum: [24h, 7d, 30d]
+ default: 7d
+ - name: scope
+ in: query
+ description: Your requests, or your team's requests. Team scope requires being a team admin.
+ schema:
+ type: string
+ enum: [self, team]
+ default: self
+ responses:
+ "200":
+ description: Usage for the selected range and scope.
+ content:
+ application/json:
+ schema:
+ $ref: "#/components/schemas/UsageResponse"
+ "400":
+ description: Invalid query parameters.
+ content:
+ application/json:
+ schema:
+ $ref: "#/components/schemas/ErrorResponse"
+ "401":
+ description: Invalid credentials.
+ content:
+ application/json:
+ schema:
+ $ref: "#/components/schemas/ErrorResponse"
+ "403":
+ description: Account suspended or team scope requested without team admin access.
+ content:
+ application/json:
+ schema:
+ $ref: "#/components/schemas/ErrorResponse"
+ "429":
+ description: Too many requests.
+ content:
+ application/json:
+ schema:
+ $ref: "#/components/schemas/ErrorResponse"
+ headers:
+ Retry-After:
+ description: Seconds to wait before retrying.
+ schema:
+ type: integer
+ minimum: 1
+ "503":
+ description: Usage reporting is temporarily unavailable.
+ content:
+ application/json:
+ schema:
+ $ref: "#/components/schemas/ErrorResponse"
+ /api/balance:
+ get:
+ summary: Balance
+ description: Check your remaining usage credits.
+ operationId: balance
+ servers:
+ - url: https://ollama.com
+ description: Ollama Cloud
+ security:
+ - bearerAuth: []
+ responses:
+ "200":
+ description: Current included and purchased balances.
+ content:
+ application/json:
+ schema:
+ $ref: "#/components/schemas/BalanceResponse"
+ "400":
+ description: Invalid query parameters.
+ content:
+ application/json:
+ schema:
+ $ref: "#/components/schemas/ErrorResponse"
+ "401":
+ description: Invalid credentials.
+ content:
+ application/json:
+ schema:
+ $ref: "#/components/schemas/ErrorResponse"
+ "403":
+ description: Account suspended.
+ content:
+ application/json:
+ schema:
+ $ref: "#/components/schemas/ErrorResponse"
+ "429":
+ description: Too many requests.
+ content:
+ application/json:
+ schema:
+ $ref: "#/components/schemas/ErrorResponse"
+ headers:
+ Retry-After:
+ description: Seconds to wait before retrying.
+ schema:
+ type: integer
+ minimum: 1
+ "503":
+ description: Balance reporting is temporarily unavailable.
+ content:
+ application/json:
+ schema:
+ $ref: "#/components/schemas/ErrorResponse"
diff --git a/server/cloud_proxy.go b/server/cloud_proxy.go
index 922df796c..51db28d0d 100644
--- a/server/cloud_proxy.go
+++ b/server/cloud_proxy.go
@@ -386,9 +386,15 @@ func signCloudProxyRequest(ctx context.Context, req *http.Request) error {
}
func buildCloudSignatureChallenge(req *http.Request, ts string) string {
- query := req.URL.Query()
- query.Set("ts", ts)
- req.URL.RawQuery = query.Encode()
+ query, err := url.ParseQuery(req.URL.RawQuery)
+ if err != nil {
+ // Preserve malformed queries for the cloud to reject. Its verifier uses
+ // the first timestamp, so prepend ours ahead of any client value.
+ req.URL.RawQuery = "ts=" + ts + "&" + req.URL.RawQuery
+ } else {
+ query.Set("ts", ts)
+ req.URL.RawQuery = query.Encode()
+ }
return fmt.Sprintf("%s,%s", req.Method, req.URL.RequestURI())
}
diff --git a/server/cloud_proxy_test.go b/server/cloud_proxy_test.go
index 950ec2bc2..05788bad4 100644
--- a/server/cloud_proxy_test.go
+++ b/server/cloud_proxy_test.go
@@ -249,6 +249,30 @@ func TestBuildCloudSignatureChallengeOverwritesExistingTimestamp(t *testing.T) {
}
}
+func TestBuildCloudSignatureChallengePreservesMalformedQuery(t *testing.T) {
+ for _, rawQuery := range []string{
+ "range=24h;",
+ "range=24h&scope=%zz",
+ "ts=999&range=24h;",
+ "ts=%zz&range=24h",
+ } {
+ t.Run(rawQuery, func(t *testing.T) {
+ req := httptest.NewRequestWithContext(t.Context(), http.MethodGet, "/api/usage?"+rawQuery, nil)
+ challenge := buildCloudSignatureChallenge(req, "123")
+ wantQuery := "ts=123&" + rawQuery
+ if got := req.URL.RawQuery; got != wantQuery {
+ t.Errorf("signed query = %q, want %q", got, wantQuery)
+ }
+ if want := "GET,/api/usage?" + wantQuery; challenge != want {
+ t.Errorf("challenge = %q, want %q", challenge, want)
+ }
+ if got := req.URL.Query().Get("ts"); got != "123" {
+ t.Errorf("verifier timestamp = %q, want 123", got)
+ }
+ })
+ }
+}
+
func TestJSONLFramingResponseWriter_SplitsCoalescedLines(t *testing.T) {
rec := &chunkRecorder{header: http.Header{}}
w := &jsonlFramingResponseWriter{ResponseWriter: rec}
diff --git a/server/routes.go b/server/routes.go
index 90ed6789a..addc6c189 100644
--- a/server/routes.go
+++ b/server/routes.go
@@ -62,6 +62,8 @@ const (
cloudErrRemoteModelDetailsUnavailable = "remote model details are unavailable"
cloudErrWebSearchUnavailable = "web search is unavailable"
cloudErrWebFetchUnavailable = "web fetch is unavailable"
+ cloudErrBalanceUnavailable = "balance is unavailable"
+ cloudErrUsageUnavailable = "usage is unavailable"
copilotChatUserAgentPrefix = "GitHubCopilotChat/"
)
@@ -2305,6 +2307,8 @@ func (s *Server) GenerateRoutes() (http.Handler, error) {
r.DELETE("/api/delete", s.DeleteHandler)
r.POST("/api/me", s.WhoamiHandler)
+ r.GET("/api/balance", s.BalanceHandler)
+ r.GET("/api/usage", s.UsageHandler)
r.POST("/api/signout", s.SignoutHandler)
// deprecated
@@ -2576,6 +2580,20 @@ func (s *Server) StatusHandler(c *gin.Context) {
})
}
+func (s *Server) BalanceHandler(c *gin.Context) {
+ c.Header("Cache-Control", "private, no-store")
+ // Use the account signed into this server, regardless of caller credentials.
+ c.Request.Header.Del("Authorization")
+ proxyCloudRequest(c, nil, cloudErrBalanceUnavailable)
+}
+
+func (s *Server) UsageHandler(c *gin.Context) {
+ c.Header("Cache-Control", "private, no-store")
+ // Use the account signed into this server, regardless of caller credentials.
+ c.Request.Header.Del("Authorization")
+ proxyCloudRequest(c, nil, cloudErrUsageUnavailable)
+}
+
func (s *Server) WebSearchExperimentalHandler(c *gin.Context) {
s.webExperimentalProxyHandler(c, "/api/web_search", cloudErrWebSearchUnavailable)
}
diff --git a/server/routes_reporting_test.go b/server/routes_reporting_test.go
new file mode 100644
index 000000000..a420ebbb5
--- /dev/null
+++ b/server/routes_reporting_test.go
@@ -0,0 +1,196 @@
+package server
+
+import (
+ "context"
+ "errors"
+ "net/http"
+ "net/http/httptest"
+ "strings"
+ "testing"
+
+ "github.com/gin-gonic/gin"
+)
+
+func TestReportingPassthrough(t *testing.T) {
+ gin.SetMode(gin.TestMode)
+ t.Setenv("OLLAMA_NO_CLOUD", "false")
+ testHome := t.TempDir()
+ setTestHome(t, testHome)
+ writeTestOllamaPrivateKey(t, testHome)
+
+ originalBaseURL, originalSigningHost := cloudProxyBaseURL, cloudProxySigningHost
+ cloudProxySigningHost = "127.0.0.1"
+ t.Cleanup(func() {
+ cloudProxyBaseURL, cloudProxySigningHost = originalBaseURL, originalSigningHost
+ })
+
+ s := &Server{}
+ router, err := s.GenerateRoutes()
+ if err != nil {
+ t.Fatal(err)
+ }
+
+ for _, tt := range []struct {
+ name string
+ path string
+ query string
+ malformed bool
+ status int
+ body string
+ retryAfter string
+ }{
+ {
+ name: "balance",
+ path: "/api/balance",
+ status: http.StatusOK,
+ body: `{"purchased":{"balance_usd":25},"future_field":true}`,
+ },
+ {
+ name: "usage",
+ path: "/api/usage",
+ query: "range=24h&scope=team",
+ status: http.StatusOK,
+ body: `{"totals":{"request_count":2},"future_field":true}`,
+ },
+ {
+ name: "upstream error",
+ path: "/api/balance",
+ status: http.StatusTooManyRequests,
+ body: `{"error":"rate limit exceeded"}`,
+ retryAfter: "60",
+ },
+ {
+ name: "malformed query",
+ path: "/api/usage",
+ query: "range=24h;",
+ malformed: true,
+ status: http.StatusBadRequest,
+ body: `{"error":"upstream rejected malformed query"}`,
+ },
+ } {
+ t.Run(tt.name, func(t *testing.T) {
+ requests := make(chan *http.Request, 1)
+ upstream := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+ requests <- r.Clone(t.Context())
+ w.Header().Set("Content-Type", "application/json")
+ if tt.retryAfter != "" {
+ w.Header().Set("Retry-After", tt.retryAfter)
+ }
+ w.WriteHeader(tt.status)
+ _, _ = w.Write([]byte(tt.body))
+ }))
+ defer upstream.Close()
+ cloudProxyBaseURL = upstream.URL
+
+ req := httptest.NewRequestWithContext(t.Context(), http.MethodGet, tt.path+"?ts=old&"+tt.query, nil)
+ req.Header.Set("Authorization", "Bearer caller-credential")
+ w := httptest.NewRecorder()
+ router.ServeHTTP(w, req)
+
+ if w.Code != tt.status || w.Body.String() != tt.body {
+ t.Fatalf("response = %d %s, want %d %s", w.Code, w.Body, tt.status, tt.body)
+ }
+ for header, want := range map[string]string{
+ "Content-Type": "application/json",
+ "Cache-Control": "private, no-store",
+ "Retry-After": tt.retryAfter,
+ } {
+ if got := w.Header().Get(header); got != want {
+ t.Errorf("%s = %q, want %q", header, got, want)
+ }
+ }
+
+ select {
+ case forwarded := <-requests:
+ if forwarded.Method != http.MethodGet || forwarded.URL.Path != tt.path {
+ t.Fatalf("forwarded request = %s %s", forwarded.Method, forwarded.URL.Path)
+ }
+ query := forwarded.URL.Query()
+ if ts := query.Get("ts"); ts == "" || ts == "old" {
+ t.Errorf("expected a fresh signing timestamp, got %q", ts)
+ }
+ if tt.malformed {
+ _, original, ok := strings.Cut(forwarded.URL.RawQuery, "&")
+ if !ok || original != req.URL.RawQuery {
+ t.Errorf("original query was changed: forwarded %q, original %q", forwarded.URL.RawQuery, req.URL.RawQuery)
+ }
+ } else {
+ query.Del("ts")
+ if got := query.Encode(); got != tt.query {
+ t.Errorf("forwarded query = %q, want %q", got, tt.query)
+ }
+ }
+ verifySignedOllamaRequest(t, forwarded)
+ default:
+ t.Fatal("reporting request was not forwarded")
+ }
+ })
+ }
+}
+
+func TestReportingLocalErrors(t *testing.T) {
+ gin.SetMode(gin.TestMode)
+ setTestHome(t, t.TempDir())
+
+ originalSignRequest := cloudProxySignRequest
+ t.Cleanup(func() {
+ cloudProxySignRequest = originalSignRequest
+ })
+
+ s := &Server{}
+ router, err := s.GenerateRoutes()
+ if err != nil {
+ t.Fatal(err)
+ }
+
+ for _, endpoint := range []string{"balance", "usage"} {
+ t.Run(endpoint, func(t *testing.T) {
+ for _, tt := range []struct {
+ name string
+ method string
+ noCloud string
+ status int
+ signCalls int
+ }{
+ {
+ name: "cloud disabled", method: http.MethodGet, noCloud: "true",
+ status: http.StatusForbidden,
+ },
+ {
+ name: "signing failed", method: http.MethodGet, noCloud: "false",
+ status: http.StatusUnauthorized, signCalls: 1,
+ },
+ {
+ name: "method not allowed", method: http.MethodPost, noCloud: "false",
+ status: http.StatusMethodNotAllowed,
+ },
+ } {
+ t.Run(tt.name, func(t *testing.T) {
+ t.Setenv("OLLAMA_NO_CLOUD", tt.noCloud)
+ signCalls := 0
+ cloudProxySignRequest = func(_ context.Context, req *http.Request) error {
+ signCalls++
+ if got := req.Header.Get("Authorization"); got != "" {
+ t.Errorf("caller credentials reached signer: %q", got)
+ }
+ return errors.New("signing failed")
+ }
+
+ req := httptest.NewRequestWithContext(t.Context(), tt.method, "/api/"+endpoint, nil)
+ req.Header.Set("Authorization", "Bearer caller-credential")
+ w := httptest.NewRecorder()
+ router.ServeHTTP(w, req)
+ if w.Code != tt.status {
+ t.Fatalf("status = %d, want %d (%s)", w.Code, tt.status, w.Body)
+ }
+ if tt.method == http.MethodGet && w.Header().Get("Cache-Control") != "private, no-store" {
+ t.Error("reporting error response must not be cached")
+ }
+ if signCalls != tt.signCalls {
+ t.Errorf("signer called %d times, want %d", signCalls, tt.signCalls)
+ }
+ })
+ }
+ })
+ }
+}