Position in SVTR
Snapshot 2026-10-29- Startup SignalsTop 25%↑34
- Foundation ModelsTop 50%
- AgentsTop 25%
- InferenceTop 50%
- AI Memory#2of 11
About
Shanghai-based large-model company building "memory science" infrastructure; its MemOS AI memory operating system unifies parametric, activation and plaintext memory.
Full tier & rationale — an SVTR Intelligence benefit
Capital graph
Company· 12
Investor· 9
Funding history
- 2026-09-07Pre-Series A+
Investors · Oriental Fortune Capital, Fortera Capital, 要弘创投, 无锡新尚资本Source · chinaventure.com.cn ↗
- 2026-07-20Pre-Series A亿元
Investors · Sense Capital, Shenzhen Capital Group, HONOR Strategic Investment, Hubble, 和玉资本Source · news.pedaily.cn ↗
Comparable companies· by market
This company
BusinessLatest roundTotal raisedDate- MemTensor
LLM memory science, AI memory operating system (MemOS) and memory-native general foundation models
Series A · 2026-09
—
United States
BusinessLatest roundTotal raisedDate- Celeris
Ultra-low-latency general-purpose diffusion-based large language model (Celeris-1), offered via an OpenAI-compatible API
— - Letta
Helps AI models remember users and conversations
Undisclosed · $10M · 2024-09
$10M - Mem0
Stores and reuses user memory to personalize AI interactions
Series A · $24M · 2025-10
$24M
China
BusinessLatest roundTotal raisedDate- 丘脑智能
Multimodal long-term memory infrastructure (Memory-as-Infrastructure) for AI agents and embodied AI, delivered via an ADK agent development kit and APIs
Seed · $7M · 2026-07
$7M - MemoraX AI
Long-term memory infrastructure for large models (intrinsic memory module + Agentic RL)
Seed · 2026-08
— - Stepfun
General-purpose large model and multimodal AI company developing the Step series of language, speech and multimodal/GUI models
Pre-IPO · $3.3B · 2026-05
$3.3B
Europe
BusinessLatest roundTotal raisedDate- H
Computer-use AI agent platform for large European enterprises. Its core Holo model series can operate existing software interfaces directly to automate business processes; the company also offers an Agents API, a Models API and the Chrome extension HoloTab.
Seed · $220M · 2024-05
$220M - Cognee
Open-source memory engine for AI agents: knowledge graph plus vector retrieval for persistent cross-session long-term memory, an ECL pipeline connecting 38+ data sources including Slack, GitHub and Linear, plus a hosted version and MCP server; customers include agent developers and organizations such as Bayer, the University of Wyoming and Knowunity.
$8M
$8M - Interloom
Enterprise memory layer and workflow orchestration platform that turns expert operational knowledge into resources callable by AI agents
Seed · $20M · 2026-03
$20M
Other / unspecified
BusinessLatest roundTotal raisedDate- UkisAI
Trains and open-sources efficient reasoning LLMs (Swift series), and provides custom AI solutions and AI training for enterprises
— - Impala AI
Infrastructure that lowers LLM inference costs for enterprises
Seed · $11M · 2025-11
$11M - Cohere
Platform company that builds general-purpose and vertical large models for enterprises
Undisclosed · $1.8B · 2025-09 · Valuation $5.5B
$1.8BValuation $5.5B
Confirmed rounds only. Matched on subsector, sector, stage and funding scale — each verifiable on the company profiles.
Related coverage
Investors that may fit
Tencent
6 Foundation Models deals in the last 24 months · 4 at a nearby stage · 6 in East China
Request intro →5Y Capital
4 Foundation Models deals in the last 24 months · 3 at a nearby stage · 4 in East China
Request intro →IDG Capital
3 Foundation Models deals in the last 24 months · 1 at a nearby stage · 3 in East China
Request intro →Shokz
3 Foundation Models deals in the last 24 months · 3 at a nearby stage · 3 in East China
Request intro →Huaqin Technology
3 Foundation Models deals in the last 24 months · 2 at a nearby stage · 3 in East China
Request intro →Qiming Venture Partners
3 Foundation Models deals in the last 24 months · 3 at a nearby stage · 3 in East China
Request intro →Ranked by each firm’s deals in this sector over the last 24 months, weighted by nearby stage and shared geography. Indicative only — no rating judgment.