bodhi-realtime-agent / RealtimeUsageModalityBreakdown

Interface: RealtimeUsageModalityBreakdown

Defined in: src/types/transport.ts:395

Optional per-modality token breakdown when the provider exposes it.

Properties

cachedAudioTokens?

optional cachedAudioTokens?: number

Defined in: src/types/transport.ts:401

---

cachedImageTokens?

optional cachedImageTokens?: number

Defined in: src/types/transport.ts:402

---

cachedTextTokens?

optional cachedTextTokens?: number

Defined in: src/types/transport.ts:400

---

cachedTokens?

optional cachedTokens?: number

Defined in: src/types/transport.ts:399

---

inputAudioTokens?

optional inputAudioTokens?: number

Defined in: src/types/transport.ts:397

---

inputImageTokens?

optional inputImageTokens?: number

Defined in: src/types/transport.ts:398

---

inputTextTokens?

optional inputTextTokens?: number

Defined in: src/types/transport.ts:396

---

outputAudioTokens?

optional outputAudioTokens?: number

Defined in: src/types/transport.ts:404

---

outputTextTokens?

optional outputTextTokens?: number

Defined in: src/types/transport.ts:403

---

reasoningTokens?

optional reasoningTokens?: number

Defined in: src/types/transport.ts:409

Reasoning tokens generated internally by reasoning-capable models (e.g. gpt-realtime-2). Hidden from the API — only the count is exposed via response.usage.output_tokens_details.reasoning_tokens. Billed as output tokens at the chosen modality rate.