Gemini Model Glossary
The Gemini Model Glossary is a collection of every major AI model from πGoogle and πDeepMind, from LaMDA and PaLM through the πGemini family, the open-weight Gemma models, and Google's image, video, and audio models, grouped by family and listed newest first. As of October 13, 2026, Google's newest and most capable model is Gemini 4 Argon, the first of the Gemini 4 family.
Glossary
Gemini 4
- πGemini 4 Argon β announced October 13, 2026: Google's new frontier model and the first of the Gemini 4 family, with a 1M-token unified input-output limit.
Gemini 3
- Gemini 3.8 Flash β
gemini-3.8-flash, released September 2, 2026: Google's newest model, tuned for long-horizon coding and autonomous agents, with a 1M-token context window and 64K output. - Gemini 3.8 Live β
gemini-3.8-live, 2026: the default Live API model for low-latency voice agents, with an Extended Thinking variant for harder voice tasks. - Gemini 3.7 Flash β
gemini-3.7-flash, released August 13, 2026: the previous Flash release, aimed at complex coding and agentic work. - Gemini 3.6 Flash β
gemini-3.6-flash, released July 21, 2026: a Flash model balancing speed and multimodal capability, launched alongside Gemini 3.5 Flash-Lite. - Gemini 3.5 Flash β
gemini-3.5-flash, released May 19, 2026 at Google I/O: generally available on day one and made the default model in the Gemini app and AI Mode in Search. - Gemini 3.1 Flash-Lite β
gemini-3.1-flash-lite, previewed March 3, 2026 and stable May 7, 2026: a low-cost model scheduled to shut down May 7, 2027. - Gemini 3.1 Pro β
gemini-3.1-pro-preview, released February 19, 2026: Google's most capable Pro model, still in preview, with Gemini 3.5 Pro not yet released. - Gemini 3 Flash β
gemini-3-flash-preview, released December 17, 2025: brought Gemini 3 Pro-level reasoning to Flash speed and became the Gemini app's default. - Gemini 3 Pro β released November 18, 2025: the first Gemini 3 model; its preview API shut down March 9, 2026 in favor of Gemini 3.1 Pro.
Gemini 2
- Gemini 2.5 Pro β
gemini-2.5-pro, introduced March 2025 and stable June 2025: Google's first "thinking" Pro model, strong on reasoning and coding. - Gemini 2.5 Flash and Flash-Lite β
gemini-2.5-flashandgemini-2.5-flash-lite, stable June and July 2025: fast, low-cost reasoning models. - Gemini 2.0 Flash β
gemini-2.0-flash, previewed December 11, 2024 and stable February 5, 2025: built for the "agentic era"; API shut down June 1, 2026. - Gemini 2.0 Flash-Lite β
gemini-2.0-flash-lite, released February 25, 2025: the cheapest Gemini 2.0 model; API shut down June 1, 2026.
Gemini 1
- Gemini 1.5 Pro β released February 15, 2024: introduced a 1M-token context window, later extended to 2M tokens.
- Gemini 1.5 Flash β released May 14, 2024: the first Flash model, a lighter version of 1.5 Pro.
- Gemini 1.0 Ultra, Pro, and Nano β announced December 6, 2023: Google's first natively multimodal models, with Nano sized for on-device use.
Predecessors
- Bard β launched March 21, 2023: Google's first public AI chatbot, renamed Gemini on February 8, 2024.
- PaLM 2 β announced May 10, 2023: the model that powered Bard before Gemini.
- PaLM β announced April 2022: a 540-billion-parameter language model.
- LaMDA β announced May 2021: the conversational model behind Bard's first version.
Gemma (Open-Weight)
- Gemma 4 β released April 2, 2026: four open models from phone-sized E2B to 31B, the first Gemma release under the Apache 2.0 license.
- Gemma 3n β released June 2025: a mobile-first Gemma model for on-device use.
- Gemma 3 β released March 12, 2025: open models built from Gemini 2.0 research.
- Gemma 2 β released June 27, 2024: the second generation of Google's open models.
- Gemma β released February 21, 2024: Google's first open-weight models, built from Gemini research.
Image
- Nano Banana 2 and Nano Banana 2 Lite β
gemini-3.1-flash-imageandgemini-3.1-flash-lite-image, released from February 26, 2026: Google's current image generation and editing models. - Nano Banana Pro β
gemini-3-pro-image, released November 20, 2025: high-end image generation built on Gemini 3 Pro. - πNano Banana β
gemini-2.5-flash-image, released August 26, 2025: the viral image-editing model; API shuts down October 2, 2026. - Imagen 4 β released June 24, 2025 in the API: Google's last standalone text-to-image model; shut down August 17, 2026.
- Imagen 3 β released August 2024: shut down in the API November 10, 2025.
Video
- Gemini Omni Flash β
gemini-omni-1.1-flash, announced May 19, 2026: Google's first any-to-any model, turning mixed text, image, audio, and video input into edited video with native audio. - Veo 3.1 and Veo 3.1 Lite β released from October 15, 2025: Google's current dedicated video generation models, in preview.
- πGoogle Veo 3 β announced May 20, 2025: the first widely available video model with native audio; API shut down June 30, 2026.
- Veo 2 β released December 16, 2024: API shut down June 30, 2026.
- Veo β announced May 14, 2024: Google's first Veo video model.
Audio and Music
- Lyria 3.5 β
lyria-3.5, released September 4, 2026: Google's flagship music model, generating full songs of up to three minutes. - Gemini 3.8 Flash TTS and Flash-Lite TTS β 2026: Google's current text-to-speech models, covering up to 130 languages.
- Gemini 3.5 Transcribe β 2026: low-latency speech-to-text with language detection and speaker diarization.
Agents and Specialized
- Antigravity Agent β
antigravity-preview-09-2026, first previewed May 19, 2026: Google's general-purpose managed agent in the Gemini API. - Gemini Deep Research and Deep Research Max β April 2026 previews: agents that plan and run multi-step research.
- Gemini 2.5 Computer Use β October 2025 preview: a model that operates user interfaces from screenshots.
- Gemini Robotics ER 2 β 2026 preview: embodied reasoning for robots, following Gemini Robotics' debut in March 2025.
- Genie 3 β announced August 5, 2025: a research world model that generates interactive 3D environments.
