Amazon Nova family

https://www.aboutamazon.com/news/aws/amazon-nova-foundation-models

Amazon's first-party foundation model family, launched at AWS re:Invent Dec-2024. Nova Pro (multimodal mid-tier), Nova Lite (multimodal cheap), Nova Micro (text-only fast), Nova Canvas (image gen), Nova Reel (video gen). Premier frontier-tier launched 2025. Substrate for AWS Bedrock-native agents and Amazon Q. Distinct from Claude / Llama / Mistral hosted via Bedrock — Nova is Amazon's own training.

At a glance

Type
AWS-native foundation model family (Nova Pro / Lite / Micro / Premier / Canvas / Reel)
Tier
T2
Created
2024-12 (Nova family launched at re:Invent); 2025 (Nova Premier)
Latest release
Nova Premier (2025); Nova Pro / Lite / Micro / Canvas / Reel (2024-12)
License
not applicable — not OSS
GitHub
not applicable — no GitHub repo
Pricing
AWS Bedrock pay-per-token: Nova Micro $0.035/$0.14 per 1M; Lite $0.06/$0.24; Pro $0.80/$3.20; Premier $2.50/$12.50
Funding
Amazon (AMZN) public; ~$2T market cap; capex $100B+ 2025 (largest among hyperscalers)

Taxonomy

storage
parametric
retrieval
parametric-recall
persistence
parametric-permanent
update
read-only
unit
weight
governance
opaque
conflict
n/a

When to use

Optimised for: AWS-native enterprise; cheap multimodal at Lite / Micro tier; integration with Amazon Q + Alexa+

Anti-fit: not for self-hosting (weights closed); AWS-locked deployment; benchmark performance trails Claude / GPT / Gemini at the frontier tier

Pros & cons

Pros

AWS distribution to massive existing enterprise base; cheapest multimodal at Lite / Micro tier (75% vendor-claimed); native integration with Amazon Q + Alexa+; FedRAMP High via AWS GovCloud.

Cons

Lags frontier tier on most benchmarks; AWS-only deployment; closed weights; Amazon AGI team newer than DeepMind / OpenAI / Anthropic; Premier tier still maturing.

Claims & capabilities

Nova Pro multimodal mid-tier 300k ctx; Nova Premier frontier 2025; trained by Amazon AGI team Seattle / Bellevue; AWS Bedrock-native pricing 75% cheaper than peers (Amazon-stated for Lite/Micro)

Technical surface

API surface
AWS Bedrock REST/SDK (Python / TS / Java / Go / .NET); SigV4 auth
Backend storage
not applicable — substrate foundation model
Deployment
Managed-only via AWS Bedrock; no self-hosting (weights closed)
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

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).

  • Google Gemini 3 family T1

    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).

  • 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.

Related systems

References (1)

  • Bedrock AgentCore (AWS) depends on at runtime — adjacent-infrastructure cell: AWS Bedrock (distribution); Amazon Q Developer / Q Business; Alexa+; Trainium/Inferentia chips

Row last verified 2026-05-14. Catalog data is CC-BY-4.0 — see how to read this.