DeepSeek R1 / V3 family
DeepSeek (Hangzhou-based, Liang Wenfeng founder) — released DeepSeek-V3 671B MoE (37B active) Dec-2024, then R1 reasoning model Jan-2025 that matched o1 at ~3% of inference cost. Caused 'DeepSeek moment' in markets Jan-27-2025 — NVIDIA dropped 17%. MIT license. Substrate for cost-conscious OSS deployers and an active reasoning-research baseline.
At a glance
- Type
- Open-weights reasoning-first frontier model family (R1 / V3 / V3.1)
- Tier
- T1
- Created
- 2023-07 (DeepSeek founded); 2024-05 (V2); 2024-12 (V3); 2025-01 (R1); 2025-08 (V3.1)
- Latest release
- V3.1 (2025-08), R1 (2025-01)
- License
- MIT (R1 + V3 weights)
- Pricing
- API: V3 $0.27/$1.10 per 1M tokens; R1 $0.55/$2.19; ~10-30x cheaper than Anthropic/OpenAI; weights free under MIT
- Funding
- Self-funded via High-Flyer (parent quant fund); no external VC disclosed; estimated <$100M training cost (vs $1B+ for peers)
Taxonomy
- storage
- parametric
- retrieval
- parametric-recall
- persistence
- parametric-permanent
- update
- read-only
- unit
- weight
- governance
- opaque
- conflict
- n/a
When to use
Optimised for: cost-efficient frontier reasoning; OSS-friendly deployment; distillation source for smaller models
Anti-fit: China data-residency for regulated Western workloads; export-control posture; less polished tool-use than Claude/GPT; multimodal less mature
Pros & cons
Pros
MIT-licensed frontier weights; ~3% the inference cost of peers; R1 matches o1 on math/reasoning; spawned a wave of cheaper distilled reasoners.
Cons
China-based — geopolitical and regulatory risk for Western enterprises; tool-use less polished; multimodal lagging; dataset / training-data transparency limited.
Claims & capabilities
R1 matches o1 on AIME/MATH at ~3% the cost; V3 671B MoE / 37B active / 128k context; MIT license (one of most permissive for frontier weights); used for distillation into smaller Qwen / Llama variants
Technical surface
- API surface
- OpenAI-compatible REST + SDKs; HuggingFace Transformers
- Backend storage
- not applicable — substrate foundation model
- Deployment
- Self-hostable (MIT weights) + hosted DeepSeek API + Together / Fireworks / HuggingFace inference
- 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.