@@ -71,10 +71,13 @@ type Transcription struct {
7171 // models `gpt-4o-transcribe` and `gpt-4o-mini-transcribe` if `logprobs` is added
7272 // to the `include` array.
7373 Logprobs []TranscriptionLogprob `json:"logprobs"`
74+ // Token usage statistics for the request.
75+ Usage TranscriptionUsageUnion `json:"usage"`
7476 // JSON contains metadata for fields, check presence with [respjson.Field.Valid].
7577 JSON struct {
7678 Text respjson.Field
7779 Logprobs respjson.Field
80+ Usage respjson.Field
7881 ExtraFields map [string ]respjson.Field
7982 raw string
8083 } `json:"-"`
@@ -109,6 +112,153 @@ func (r *TranscriptionLogprob) UnmarshalJSON(data []byte) error {
109112 return apijson .UnmarshalRoot (data , r )
110113}
111114
115+ // TranscriptionUsageUnion contains all possible properties and values from
116+ // [TranscriptionUsageTokens], [TranscriptionUsageDuration].
117+ //
118+ // Use the [TranscriptionUsageUnion.AsAny] method to switch on the variant.
119+ //
120+ // Use the methods beginning with 'As' to cast the union to one of its variants.
121+ type TranscriptionUsageUnion struct {
122+ // This field is from variant [TranscriptionUsageTokens].
123+ InputTokens int64 `json:"input_tokens"`
124+ // This field is from variant [TranscriptionUsageTokens].
125+ OutputTokens int64 `json:"output_tokens"`
126+ // This field is from variant [TranscriptionUsageTokens].
127+ TotalTokens int64 `json:"total_tokens"`
128+ // Any of "tokens", "duration".
129+ Type string `json:"type"`
130+ // This field is from variant [TranscriptionUsageTokens].
131+ InputTokenDetails TranscriptionUsageTokensInputTokenDetails `json:"input_token_details"`
132+ // This field is from variant [TranscriptionUsageDuration].
133+ Duration float64 `json:"duration"`
134+ JSON struct {
135+ InputTokens respjson.Field
136+ OutputTokens respjson.Field
137+ TotalTokens respjson.Field
138+ Type respjson.Field
139+ InputTokenDetails respjson.Field
140+ Duration respjson.Field
141+ raw string
142+ } `json:"-"`
143+ }
144+
145+ // anyTranscriptionUsage is implemented by each variant of
146+ // [TranscriptionUsageUnion] to add type safety for the return type of
147+ // [TranscriptionUsageUnion.AsAny]
148+ type anyTranscriptionUsage interface {
149+ implTranscriptionUsageUnion ()
150+ }
151+
152+ func (TranscriptionUsageTokens ) implTranscriptionUsageUnion () {}
153+ func (TranscriptionUsageDuration ) implTranscriptionUsageUnion () {}
154+
155+ // Use the following switch statement to find the correct variant
156+ //
157+ // switch variant := TranscriptionUsageUnion.AsAny().(type) {
158+ // case openai.TranscriptionUsageTokens:
159+ // case openai.TranscriptionUsageDuration:
160+ // default:
161+ // fmt.Errorf("no variant present")
162+ // }
163+ func (u TranscriptionUsageUnion ) AsAny () anyTranscriptionUsage {
164+ switch u .Type {
165+ case "tokens" :
166+ return u .AsTokens ()
167+ case "duration" :
168+ return u .AsDuration ()
169+ }
170+ return nil
171+ }
172+
173+ func (u TranscriptionUsageUnion ) AsTokens () (v TranscriptionUsageTokens ) {
174+ apijson .UnmarshalRoot (json .RawMessage (u .JSON .raw ), & v )
175+ return
176+ }
177+
178+ func (u TranscriptionUsageUnion ) AsDuration () (v TranscriptionUsageDuration ) {
179+ apijson .UnmarshalRoot (json .RawMessage (u .JSON .raw ), & v )
180+ return
181+ }
182+
183+ // Returns the unmodified JSON received from the API
184+ func (u TranscriptionUsageUnion ) RawJSON () string { return u .JSON .raw }
185+
186+ func (r * TranscriptionUsageUnion ) UnmarshalJSON (data []byte ) error {
187+ return apijson .UnmarshalRoot (data , r )
188+ }
189+
190+ // Usage statistics for models billed by token usage.
191+ type TranscriptionUsageTokens struct {
192+ // Number of input tokens billed for this request.
193+ InputTokens int64 `json:"input_tokens,required"`
194+ // Number of output tokens generated.
195+ OutputTokens int64 `json:"output_tokens,required"`
196+ // Total number of tokens used (input + output).
197+ TotalTokens int64 `json:"total_tokens,required"`
198+ // The type of the usage object. Always `tokens` for this variant.
199+ Type constant.Tokens `json:"type,required"`
200+ // Details about the input tokens billed for this request.
201+ InputTokenDetails TranscriptionUsageTokensInputTokenDetails `json:"input_token_details"`
202+ // JSON contains metadata for fields, check presence with [respjson.Field.Valid].
203+ JSON struct {
204+ InputTokens respjson.Field
205+ OutputTokens respjson.Field
206+ TotalTokens respjson.Field
207+ Type respjson.Field
208+ InputTokenDetails respjson.Field
209+ ExtraFields map [string ]respjson.Field
210+ raw string
211+ } `json:"-"`
212+ }
213+
214+ // Returns the unmodified JSON received from the API
215+ func (r TranscriptionUsageTokens ) RawJSON () string { return r .JSON .raw }
216+ func (r * TranscriptionUsageTokens ) UnmarshalJSON (data []byte ) error {
217+ return apijson .UnmarshalRoot (data , r )
218+ }
219+
220+ // Details about the input tokens billed for this request.
221+ type TranscriptionUsageTokensInputTokenDetails struct {
222+ // Number of audio tokens billed for this request.
223+ AudioTokens int64 `json:"audio_tokens"`
224+ // Number of text tokens billed for this request.
225+ TextTokens int64 `json:"text_tokens"`
226+ // JSON contains metadata for fields, check presence with [respjson.Field.Valid].
227+ JSON struct {
228+ AudioTokens respjson.Field
229+ TextTokens respjson.Field
230+ ExtraFields map [string ]respjson.Field
231+ raw string
232+ } `json:"-"`
233+ }
234+
235+ // Returns the unmodified JSON received from the API
236+ func (r TranscriptionUsageTokensInputTokenDetails ) RawJSON () string { return r .JSON .raw }
237+ func (r * TranscriptionUsageTokensInputTokenDetails ) UnmarshalJSON (data []byte ) error {
238+ return apijson .UnmarshalRoot (data , r )
239+ }
240+
241+ // Usage statistics for models billed by audio input duration.
242+ type TranscriptionUsageDuration struct {
243+ // Duration of the input audio in seconds.
244+ Duration float64 `json:"duration,required"`
245+ // The type of the usage object. Always `duration` for this variant.
246+ Type constant.Duration `json:"type,required"`
247+ // JSON contains metadata for fields, check presence with [respjson.Field.Valid].
248+ JSON struct {
249+ Duration respjson.Field
250+ Type respjson.Field
251+ ExtraFields map [string ]respjson.Field
252+ raw string
253+ } `json:"-"`
254+ }
255+
256+ // Returns the unmodified JSON received from the API
257+ func (r TranscriptionUsageDuration ) RawJSON () string { return r .JSON .raw }
258+ func (r * TranscriptionUsageDuration ) UnmarshalJSON (data []byte ) error {
259+ return apijson .UnmarshalRoot (data , r )
260+ }
261+
112262type TranscriptionInclude string
113263
114264const (
@@ -131,11 +281,14 @@ type TranscriptionStreamEventUnion struct {
131281 Logprobs TranscriptionStreamEventUnionLogprobs `json:"logprobs"`
132282 // This field is from variant [TranscriptionTextDoneEvent].
133283 Text string `json:"text"`
134- JSON struct {
284+ // This field is from variant [TranscriptionTextDoneEvent].
285+ Usage TranscriptionTextDoneEventUsage `json:"usage"`
286+ JSON struct {
135287 Delta respjson.Field
136288 Type respjson.Field
137289 Logprobs respjson.Field
138290 Text respjson.Field
291+ Usage respjson.Field
139292 raw string
140293 } `json:"-"`
141294}
@@ -279,11 +432,14 @@ type TranscriptionTextDoneEvent struct {
279432 // [create a transcription](https://platform.openai.com/docs/api-reference/audio/create-transcription)
280433 // with the `include[]` parameter set to `logprobs`.
281434 Logprobs []TranscriptionTextDoneEventLogprob `json:"logprobs"`
435+ // Usage statistics for models billed by token usage.
436+ Usage TranscriptionTextDoneEventUsage `json:"usage"`
282437 // JSON contains metadata for fields, check presence with [respjson.Field.Valid].
283438 JSON struct {
284439 Text respjson.Field
285440 Type respjson.Field
286441 Logprobs respjson.Field
442+ Usage respjson.Field
287443 ExtraFields map [string ]respjson.Field
288444 raw string
289445 } `json:"-"`
@@ -318,6 +474,57 @@ func (r *TranscriptionTextDoneEventLogprob) UnmarshalJSON(data []byte) error {
318474 return apijson .UnmarshalRoot (data , r )
319475}
320476
477+ // Usage statistics for models billed by token usage.
478+ type TranscriptionTextDoneEventUsage struct {
479+ // Number of input tokens billed for this request.
480+ InputTokens int64 `json:"input_tokens,required"`
481+ // Number of output tokens generated.
482+ OutputTokens int64 `json:"output_tokens,required"`
483+ // Total number of tokens used (input + output).
484+ TotalTokens int64 `json:"total_tokens,required"`
485+ // The type of the usage object. Always `tokens` for this variant.
486+ Type constant.Tokens `json:"type,required"`
487+ // Details about the input tokens billed for this request.
488+ InputTokenDetails TranscriptionTextDoneEventUsageInputTokenDetails `json:"input_token_details"`
489+ // JSON contains metadata for fields, check presence with [respjson.Field.Valid].
490+ JSON struct {
491+ InputTokens respjson.Field
492+ OutputTokens respjson.Field
493+ TotalTokens respjson.Field
494+ Type respjson.Field
495+ InputTokenDetails respjson.Field
496+ ExtraFields map [string ]respjson.Field
497+ raw string
498+ } `json:"-"`
499+ }
500+
501+ // Returns the unmodified JSON received from the API
502+ func (r TranscriptionTextDoneEventUsage ) RawJSON () string { return r .JSON .raw }
503+ func (r * TranscriptionTextDoneEventUsage ) UnmarshalJSON (data []byte ) error {
504+ return apijson .UnmarshalRoot (data , r )
505+ }
506+
507+ // Details about the input tokens billed for this request.
508+ type TranscriptionTextDoneEventUsageInputTokenDetails struct {
509+ // Number of audio tokens billed for this request.
510+ AudioTokens int64 `json:"audio_tokens"`
511+ // Number of text tokens billed for this request.
512+ TextTokens int64 `json:"text_tokens"`
513+ // JSON contains metadata for fields, check presence with [respjson.Field.Valid].
514+ JSON struct {
515+ AudioTokens respjson.Field
516+ TextTokens respjson.Field
517+ ExtraFields map [string ]respjson.Field
518+ raw string
519+ } `json:"-"`
520+ }
521+
522+ // Returns the unmodified JSON received from the API
523+ func (r TranscriptionTextDoneEventUsageInputTokenDetails ) RawJSON () string { return r .JSON .raw }
524+ func (r * TranscriptionTextDoneEventUsageInputTokenDetails ) UnmarshalJSON (data []byte ) error {
525+ return apijson .UnmarshalRoot (data , r )
526+ }
527+
321528type AudioTranscriptionNewParams struct {
322529 // The audio file object (not file name) to transcribe, in one of these formats:
323530 // flac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, or webm.
![stainless-app[bot]](https://avatars.githubusercontent.com/in/378072?v=4&size=40){kind=avatar}
0 commit comments