← All models

Gemini 3.5 Live Translate

Published Publisher Googlegemini-3.5-live-translate-preview

Positioning and release

Gemini 3.5 Live Translate is a low-latency audio model optimized for live speech-to-speech translation. Google announced it on June 9, 2026, with a developer public preview through the Gemini Live API and Google AI Studio. (blog.google)

Capabilities and uses

  • Automatically detects more than 70 languages and processes speech continuously, rather than waiting for the speaker to finish. Google says translated audio typically follows the speaker by a few seconds and preserves aspects of intonation, pace and pitch. (blog.google)
  • Accepts speech audio and produces translated audio and transcript text. The documented input and output limits are 131,072 and 65,536 tokens, respectively. (aistudio.google.com)
  • Intended uses include live interpretation for multilingual calls, meetings, lessons and broadcasts. At announcement, Google also described rollout in Google Translate on Android and iOS and a private preview for selected Google Meet business customers. (blog.google)
  • Generated audio carries a SynthID watermark. (blog.google)

Limitations

This is a preview model. Its documented capabilities do not include context caching, Batch API, function calling, code execution, structured output or thinking. (aistudio.google.com)

Google notes that voices can shift after long pauses or become inconsistent during rapid multi-speaker sessions. Language detection can struggle with non-native accents, similar languages and rapid language switches; background audio may not always be filtered out. (storage.googleapis.com)

Pricing

For the Gemini Developer API standard tier, the official USD pricing checked on November 16, 2026 is: (aistudio.google.com)

Tier Audio input Audio output
Free Free of charge Free of charge
Paid USD 3.50 per 1 million tokens; approximately USD 0.0053 per minute USD 21.00 per 1 million tokens; approximately USD 0.0315 per minute

Billing is based on total input and output audio token consumption, at 25 tokens per second of audio. Google gives an approximate combined effective rate of USD 0.0368 per minute. The free tier is subject to usage limits; Google states that free-tier content may be used to improve its products, whereas paid-tier content is not. These are Developer API prices, not a quote for other platforms. (ai.google.dev)

Official sources