API Release Notes

Google released antigravity-preview-09-2026, replacing and deprecating antigravity-preview-05-2026, which shuts down on October 5, 2026. Remote-sandbox users reading only output_text or model_output need just the agent string updated, but local tool runners and function_call parsers must switch to new built-in tools including write_to_file, replace_file_content, view_file, list_dir, find_by_name, and grep_search; shell execution and web search are unchanged.

Two audio-to-audio models for real-time voice applications on the Live API are generally available: gemini-3.8-live for low-latency voice agents and real-time dialogue with interleaved reasoning and default asynchronous function calling, and gemini-3.8-live-extended-thinking for background reasoning during live audio interactions.

Google released Lyria 3.5 in public preview, the next generation of its music generation model. It generates full-length songs with improved musical coherence, natural vocals, and fine-grained duration and structural control, producing 44.1 kHz stereo audio from text and image inputs.

Lyria 3.5 is available in public preview, generating full-length songs with improved musical coherence, natural vocals, and fine-grained duration and structural control. The model accepts text and image inputs and outputs high-fidelity 44.1 kHz stereo audio.

Released gemini-3.8-flash as generally available, positioned as the most intelligent Flash model, built for long-horizon software engineering, autonomous agents, and complex enterprise workflows.

Gemini 3.8 Flash is now generally available, positioned as the most intelligent Flash model for long-horizon software engineering, autonomous agents, and complex enterprise workflows.

Gemini 3.7 Flash, 3.6 Flash, and 3.5 Flash-Lite now support agentic video understanding through the Interactions and GenerateContent APIs. Instead of processing entire timelines statically, the model dynamically requests transcripts, frames, or audio tracks on demand, cutting token usage by up to 88% on long-form content.

Agentic video understanding is now available for Gemini 3.7 Flash, 3.6 Flash, and 3.5 Flash-Lite across the Interactions and GenerateContent APIs. The model dynamically navigates video timelines, requesting transcripts, frames, or audio tracks on demand, using up to 88% fewer tokens for long-form content compared to static processing.

Gemini Omni Flash is generally available as gemini-omni-1.1-flash, adding video extension, first+last frame interpolation, and resolution control up to 4K. The gemini-omni-flash-preview endpoint is deprecated and will be removed on September 30, 2026.

Gemini Omni Flash (gemini-omni-1.1-flash) is now generally available as a fast, conversational video generation and editing model, supporting video extension, first/last-frame interpolation, and a new resolution parameter with 360p, 720p (default), 1080p, and 4K outputs. The gemini-omni-flash-preview endpoint is deprecated and will be retired on September 30, 2026.

Two speech-to-text models are now generally available: Gemini 3.5 Transcribe for high-accuracy non-streaming transcription with utterance-based language detection across 85+ languages, and Gemini 3.5 Transcribe Live for low-latency bidirectional streaming via the Live API.

Two dedicated speech-to-text models are generally available: Gemini 3.5 Transcribe, a non-streaming model with utterance-based language detection across 85+ languages, speaker diarization, word-level timestamps, and custom vocabulary biasing; and Gemini 3.5 Transcribe Live, a bidirectional streaming model over WebSockets with interim and finalized transcription events, Smart transcription mode, and multiple VAD strategies.

Gemini 3.7 Flash is now generally available as gemini-3.7-flash, with substantial improvements across software engineering, web development, and agentic workflows seats at an introductory price through December 31, 2026.

Gemini 3.7 Flash (gemini-3.7-flash) is now generally available, bringing substantial improvements across software engineering, web development, and agentic workflows. Introductory pricing applies through December 31, 2026.

Two new embodied reasoning model endpoints — gemini-robotics-er-2-preview and gemini-robotics-er-2-streaming-preview — are now available in public preview. The previous gemini-robotics-er-1.6-preview model will be shut down on August 31, 2026.

Gemini Robotics ER 2 in public preview: Released two new embodied reasoning model endpoints for robotics:

• gemini-robotics-er-2-preview: Advanced spatial reasoning, agentic code execution, multi-step tool orchestration, video moment finding, progress classification, and…

Gemini 3.6 Flash and Gemini 3.5 Flash-Lite generally available (GA): Released stable, production-ready versions of our latest 3.x Flash models:

• Gemini 3.6 Flash (gemini-3.6-flash): Features improved token efficiency and code/agentic planning capabilities at a lower…

Stable, production-ready versions of Gemini 3.6 Flash and Gemini 3.5 Flash-Lite are now available, with 3.6 Flash offering improved token efficiency and code/agentic planning at a lower price. The sampling parameters temperature, top_p, and top_k are deprecated.

Developer logs support for the Interactions API: logs for supported Interactions API calls are now viewable in the AI Studio dashboard.

Logs for supported Interactions API calls are now viewable in the AI Studio dashboard.