Google Gemini 3 family
https://blog.google/technology/google-deepmind/gemini-3/
Google DeepMind's Gemini 3 frontier model family — Gemini 3 Pro (top tier), 3 Flash (fast), 3 Nano (on-device). Substrate for Google Workspace AI, Vertex AI Agent Platform, Gemini Code Assist, Project Mariner, and Project Astra. Multimodal-native; 1M-10M context window. Largest context window of any frontier model. Used by Gemini Enterprise Agent Platform Memory Bank (separate row).
At a glance
- Type
- Frontier foundation model family (Gemini 3 Pro / Flash / Nano)
- Tier
- T1
- Created
- 2023-12 (Gemini 1.0); 2024-02 (1.5 Pro); 2024-12 (2.0); 2025-03 (2.5 Pro); 2025-12 (Gemini 3)
- Latest release
- Gemini 3 Pro (2025-12), Gemini 3 Flash, Gemini 3 Nano
- License
- not applicable — not OSS
- GitHub
- not applicable — no GitHub repo
- Pricing
- Pay-per-token (Gemini 3 Pro $1.25/$10 per 1M; Flash $0.30/$2.50); Gemini Advanced ($20/mo consumer); Gemini Enterprise pricing via GCP
- Funding
- Alphabet (GOOGL) public; ~$2T+ market cap; Google AI capex $75B+ 2025
Taxonomy
- storage
- parametric
- retrieval
- parametric-recall
- persistence
- parametric-permanent
- update
- read-only
- unit
- weight
- governance
- opaque
- conflict
- n/a
When to use
Optimised for: multimodal reasoning, ultra-long context (10M tokens), GCP-native enterprise deployments
Anti-fit: Gemini Apps consumer privacy concerns; Vertex consumption pricing complex; weights closed (Gemma is separate open-weights family)
Pros & cons
Pros
Largest context window (1M-10M tokens) of frontier tier; native multimodal (single architecture); GCP enterprise distribution; FedRAMP High; Gemma open-weights companion family.
Cons
Smaller LMSYS Arena share than Claude/GPT; consumer Gemini apps less polished than ChatGPT; pricing complexity at Vertex; lags Claude on agentic coding benchmarks.
Claims & capabilities
Gemini 3 Pro 2025-12 launch; 1M-10M token context (largest of frontier tier); multimodal-native (text+image+video+audio+code); state-of-art on long-context retrieval benchmarks
Technical surface
- API surface
- REST + gRPC + SDK (Python, Node, Go, Java, .NET, Dart); Google AI Studio + Vertex AI
- Backend storage
- not applicable — substrate foundation model
- Deployment
- Managed cloud (Google AI Studio + GCP Vertex AI); on-prem via Gemini-Distributed-Cloud + Gemma open-weights
- Embedding model
- not applicable — not a memory product
- Multi-tenancy
- not applicable — substrate foundation model
- MCP
- not applicable — substrate foundation model
- A2A
- not applicable — substrate foundation model
- OpenTelemetry
- not applicable — substrate foundation model
Compare Google Gemini 3 family with…
Similar systems
Other foundation models (substrate reference) in the catalog, ranked by inbound references.
- Anthropic Claude (foundation models) T1
Claude family of frontier foundation models — Sonnet 4 (workhorse), Opus 4.5 (frontier capability), Haiku 4.5 (fast/cheap). Frontier model used as substrate by Claude Code, Anthropic Memory tool, Cline, Aider, Continue.dev, Cursor, Goose, and dozens of other catalog entries. Memory in Claude is parametric (in weights) — discrete episodic memory lives outside the model in the Claude Memory tool (see Platform-provider memory section).
- OpenAI GPT family (GPT-5 / GPT-4o / o3 / o4) T1
OpenAI's foundation model family — GPT-5 (2025-08 flagship), GPT-4o (multimodal workhorse), o3/o4 reasoning series. Substrate for ChatGPT, the OpenAI API, Microsoft Copilot, Azure OpenAI, and many downstream agents in this catalog (Devin, Codex CLI, Operator, Cursor mode). Parametric memory in weights; discrete user memory ships in ChatGPT Memory feature (separate row).
- Mistral Large 2 / Mixtral family T1
French frontier-model lab (Paris). Family includes Mistral Large 2 (123B dense, Jul-2024), Mixtral 8x22B (open-weights MoE), Mistral Small 3 (24B, open-weights Jan-2025), Codestral (code specialist). EU-headquartered alternative to US/China model labs; GDPR-friendly data residency. Distinct from the Mistral Le Chat memory feature (already in catalog) — this row covers the model family itself as substrate.
- Meta Llama 4 family T1
Meta's Llama 4 family (released 2025-04) — Scout (109B MoE, 17B active), Maverick (400B MoE, 17B active), Behemoth (2T MoE, 288B active; preview). First Llama generation to use mixture-of-experts and to be multimodal-native. Substrate for many OSS agents (Open-Interpreter, Ollama-served local agents, vLLM-served enterprise inference). Llama community license (commercial use permitted under 700M MAU threshold).
- xAI Grok 4 T1
Elon Musk's xAI frontier model — Grok 4 launched 2025-07. Trained on Colossus (100k+ GPU supercluster in Memphis). Deeply integrated with X/Twitter (real-time tweet data access). Grok 4 Heavy variant for premium tier. No frontier-tier open-weights — Grok 1 (314B MoE) is the only weight release.
- 01.AI Yi family T2
Kai-Fu Lee's 01.AI (founded 2023, Beijing). Yi family open-weights — Yi-34B (2023), Yi-Large (proprietary 2024), Yi-Lightning (2024-10, frontier-tier, matched GPT-4o on LMSYS chatbot arena). Apache 2.0 (Yi-34B and base sizes). Important second Chinese open-weights option alongside DeepSeek + Qwen.
Related systems
Referenced by (10)
- Apptronik depends on at runtime — adjacent-infrastructure cell: Google DeepMind Gemini Robotics; Mercedes manufacturing
- Backboard depends on at runtime — adjacent-infrastructure cell: BYO LLM (OpenAI/Anthropic/Gemini/Ollama keys)
- Browser Use depends on at runtime — adjacent-infrastructure cell: Playwright; Anthropic / OpenAI / Gemini APIs; Browserbase / Hyperbrowser substrates
- Gemini Enterprise Agent Platform Memory Bank (rebrand of Vertex AI Memory Bank) depends on at runtime — adjacent-infrastructure cell: requires GCP (Vertex AI / Gemini Enterprise); on-prem option via open-source Gemma 4; managed MCP servers; Pub/Sub + Cloud Trace + IAM
- GitHub Copilot (Agent Mode) depends on at runtime — adjacent-infrastructure cell: Azure OpenAI + Anthropic Claude + Google Gemini (multi-model 2024+)
- Google DeepMind Gemini Robotics depends on at runtime — ar-2025. VLA models built on Gemini 2.0 backbone. Reference platform: Apptronik Apollo. The DeepMind line of robotics-F
- MultiOn depends on at runtime — adjacent-infrastructure cell: Playwright; Chromium-headless infrastructure; computer-use models (Claude / GPT-5 / Gemini)
- Phidata / Agno depends on at runtime — adjacent-infrastructure cell: HuggingFace Transformers; OpenAI / Anthropic / Gemini SDKs; YC W22 alumni network
- Project Mariner depends on at runtime — ental browser agent built on Gemini 2.0 — Chrome extension that takes actions on the user's behalf.
- Skyvern depends on at runtime — adjacent-infrastructure cell: Playwright; Postgres; Anthropic/OpenAI/Gemini APIs