Start free. Scale to millions of entries. Replace your entire AI memory stack for less than one Pinecone invoice.
Every plan includes all five memory types. Limits scale with the plan.
| Hobby | Builder | Pro | Scale | |
|---|---|---|---|---|
| Memory entries | 10K | 500K | 5M | Unlimited |
| API calls/month | 50K | 2M | 20M | Unlimited |
| Session memory | ✓ | ✓ | ✓ | ✓ |
| Episodic memory | ✓ | ✓ | ✓ | ✓ |
| Semantic memory | ✓ | ✓ | ✓ | ✓ |
| Persistent state | ✓ | ✓ | ✓ | ✓ |
| Temporal memory | ✓ | ✓ | ✓ | ✓ |
| MCP server | ✓ | ✓ | ✓ | ✓ |
| Time-travel | — | 30 days | Unlimited | Unlimited |
| Custom embeddings | — | — | ✓ | ✓ |
| SSO / SAML | — | — | — | ✓ |
| Uptime SLA | — | — | 99.9% | 99.99% |
| Support | Community | Priority | Dedicated engineer |
What Pro costs vs. assembling the same capabilities yourself.
One API. Zero glue code. Zero ETL. No operational overhead.
Yes. The Hobby plan is free forever — 10,000 memory entries and 50,000 API calls per month. No credit card required to start.
Each stored item — a session turn, episodic record, semantic fact, persistent state value, or temporal entry — counts as one memory entry regardless of size.
Every request to memory.store(), memory.recall(), memory.rewind(), or memory.ingest() counts as one API call.
No. Embedding generation for semantic memory is included in all plans. We run Workers AI at no additional cost to you.
Yes, on Pro and Scale. Pass your own vectors to the ingest() API instead of having Memonic generate them. Useful for domain-specific or fine-tuned models.
Plans are billed monthly. Usage resets on your billing date. You can upgrade or downgrade at any time — changes take effect immediately with prorated billing.