NAVER HyperCLOVA X THINK alternatives

Reasoning-focused successor to HyperCLOVA X. Pre-trained on ~6T high-quality Korean + English tokens. 128K context via rotary position embeddings + grouped-query attention. Sovereign reasoning LLM for Korean enterprises and government. Full NAVER HyperCLOVA X THINK record →

  1. 1 MemoraX AI Dedicated memory layers

    also in Dedicated memory layers · same storage (parametric) · same retrieval (parametric-recall)

    OSS ✓ Self-host ✓ Managed ? MCP ? released 2026-10-21 MIT (MemoraX Code client); hosted memory… Compare with NAVER HyperCLOVA X THINK →

  2. 2 NVIDIA Isaac GR00T N1.7 Vertical / domain-specific AI memory

    same storage (kv-cache) · same retrieval (attention) · also offered managed

    OSS ✓ Self-host ✓ Managed ✓ MCP ? ★ 8.2k released 2026-04-18 Apache-2.0 (code); model weights under NVIDIA… Compare with NAVER HyperCLOVA X THINK →

  3. 3 Wayve GAIA-2 / GAIA-3 Vertical / domain-specific AI memory

    same storage (kv-cache) · same retrieval (attention) · same memory unit (kv-token)

    OSS ? Self-host — Managed — MCP ? Compare with NAVER HyperCLOVA X THINK →

  4. 4 RAFT Retrieval-as-memory hybrids

    same storage (parametric) · same retrieval (parametric-recall)

    OSS ✓ Self-host ✓ Managed ? MCP ? ★ 13k Apache-2.0 Compare with NAVER HyperCLOVA X THINK →

  5. 5 Figure Helix Vertical / domain-specific AI memory

    same retrieval (attention) · same memory unit (kv-token) · also offered managed

    OSS ? Self-host — Managed ✓ MCP ? Compare with NAVER HyperCLOVA X THINK →

  6. 6 Tesla FSD V13 — Occupancy + 4D World (now FSD v14) Vertical / domain-specific AI memory ⚠ renamed — now FSD v14

    same storage (kv-cache) · same retrieval (attention)

    OSS ? Self-host — Managed — MCP ? released 2026-07-21 Compare with NAVER HyperCLOVA X THINK →

  7. 7 Mem0 Dedicated memory layers

    also in Dedicated memory layers · also offered managed

    OSS ✓ Self-host ✓ Managed ✓ MCP ✓ ★ 67k released 2026-09-25 Apache-2.0 Compare with NAVER HyperCLOVA X THINK →

  8. 8 Hindsight (Vectorize) Dedicated memory layers

    also in Dedicated memory layers · also offered managed

    OSS ✓ Self-host ✓ Managed ✓ MCP ✓ ★ 45k released 2026-10-21 MIT Compare with NAVER HyperCLOVA X THINK →

Ranked by shared memory design (storage, retrieval, conflict handling, memory unit), same section and shared must-haves (open source, self-hosting, MCP), with adoption as a tie-breaker. ✓ confirmed · — no · ? not in the catalog yet. Want picks for your own requirements? Use the chooser. Checking a whole project? Check your stack.