Transparent pricing. Graph memory at every tier.
No credit math. No graph gates. No token surprises. Pick a tier and build.
Explorer
Build your first memory-enabled agents. No card, no clock, no catch.
Forever free
- ✓3 agents
- ✓50,000 memory ops/month
- ✓100MB storage
- ✓All 4 memory types (episodic, semantic, procedural, working)
- ✓Knowledge graph (entity triples, LadybugDB-native)
- ✓Pheromone-guided retrieval
- ✓Basic usage analytics
- ✓Community support (Discord)
- ✓Hard cap at limit — no surprise charges
- ✗A* semantic pathfinding
- ✗Shared memory spaces
- ✗Active Inference agents
Builder
Serious prototypes. Team projects. The same price as Mem0, but with graph.
/month · or $190/year (save 2 months)
- ✓25 agents
- ✓500,000 memory ops/month
- ✓5GB storage
- ✓All 4 memory types
- ✓Knowledge graph
- ✓Pheromone-guided retrieval
- ✓A* semantic pathfinding
- ✓3 shared memory spaces (VC-gated access)
- ✓Full usage analytics
- ✓Email support
- ✓99.5% uptime SLA
- ✓Overage: $0.005 per 1K ops
Pro
Unlimited agents. Active Inference. $150/month less than Mem0.
/month · or $990/year (save 2 months)
- ✓Unlimited agents
- ✓5,000,000 memory ops/month
- ✓50GB storage
- ✓All 4 memory types
- ✓Knowledge graph
- ✓Pheromone-guided retrieval
- ✓A* semantic pathfinding
- ✓Unlimited shared memory spaces
- ✓Active Inference memory agents (5 archetypes)
- ✓Cross-agent queries
- ✓Full analytics + data export
- ✓Priority support (< 4hr response)
- ✓99.9% uptime SLA
- ✓Overage: $0.003 per 1K ops
Enterprise
Your agents. Your cloud. Your compliance requirements.
Starting ~$500/mo · Typical response within 1 business day
- ✓Everything in Pro
- ✓Custom agent and ops limits
- ✓99.99% uptime SLA
- ✓SOC 2 Type II certification
- ✓HIPAA BAA available
- ✓Bring Your Own Cloud (GCP BYOC)
- ✓On-premises deployment
- ✓Dedicated Customer Success Manager
- ✓Custom SLAs and billing terms
- ✓ASI Chain enterprise identity (DID/VC)
Every Tier. LadybugDB Graph. Zero LLM. Always.
Every Agentverse Memory tier includes our LadybugDB knowledge graph with zero LLM extraction. No LLM at write time (0.035ms p95 in-process engine write op, am-local). Flat cost regardless of ingestion volume. No token surprises.
| Agentverse Memory | Mem0 | RetainDB | |
|---|---|---|---|
| Graph on free tier | ✅ $0/mo | ❌ No free graph | ❌ $20/mo required |
| Graph entry price | $0/mo | $249/mo | $20/mo |
| Write architecture | Zero-LLM (TF-IDF + LadybugDB) | LLM extraction | LLM extraction |
| Write-time LLM penalty | None · 0.035ms in-process engine | ~2.59s (LLM extraction) | ~2–3s est. (LLM extraction) |
Mem0 latency per LLM extraction architecture. RetainDB latency estimated from LLM-at-write pattern; not independently verified.
How does this compare?
Transparent side-by-side. No cherry-picking.
vs Mem0
Mem0 is the market leader in agent memory — they've shipped a solid product and deserve credit. But they charge $249/month to unlock graph memory, leaving developers on the $19/month tier with vector search only. Agentverse Memory includes full knowledge graph support at every tier, including free.
vs RetainDB
RetainDB offers graph from $20/month — a real improvement over Mem0's $249 gate. But RetainDB uses LLM extraction at write time. Every remember() call invokes an LLM: 2–3s latency, token costs that scale with ingestion. If graph memory is worth having, it's worth having fast.
vs Zep
Zep's temporal knowledge graph (Graphiti) is genuinely interesting. But their credit system — 350-byte episodes, fractional credit consumption, auto-topup anxiety — adds cognitive overhead. Agentverse Memory uses flat ops/month: every read, write, or query counts as one operation. Simple math, no surprises.
Full feature comparison
| Feature | Explorer | Builder | Pro ⭐ | Enterprise |
|---|---|---|---|---|
| Knowledge graph | ✓ | ✓ | ✓ | ✓ |
| Pheromone retrieval | ✓ | ✓ | ✓ | ✓ |
| Episodic memory | ✓ | ✓ | ✓ | ✓ |
| Semantic memory | ✓ | ✓ | ✓ | ✓ |
| Procedural memory | ✓ | ✓ | ✓ | ✓ |
| Working memory | ✓ | ✓ | ✓ | ✓ |
| A* semantic pathfinding | — | ✓ | ✓ | ✓ |
| Shared memory spaces | — | 3 spaces | Unlimited | Custom |
| Active Inference agents | — | — | 5 archetypes | ✓ |
| Cross-agent queries | — | — | ✓ | ✓ |
| Analytics | Basic | Full | Full + export | Custom |
| Support | Discord | Priority <4hr | Dedicated CSM | |
| Uptime SLA | None | 99.5% | 99.9% | 99.99% |
| SOC 2 Type II | — | — | — | ✓ |
| BYOC / On-prem | — | — | — | ✓ |
| Price | Free | $19/mo | $99/mo | Custom |
Frequently asked questions
What counts as a memory operation?
Any read (recall, search, traverse), write (store, update, forget), or graph query counts as one operation. Bulk imports count as one op per document. We don't charge by byte or token.
What happens when I hit my limit on Explorer?
Hard cap — we stop accepting new operations but never delete your data. Upgrade to Builder to continue. No surprise charges.
Do you make LLM calls when I store a memory?
Never. GraphPalace extracts entities and relationships using TF-IDF — a classical information retrieval technique. No LLM at write time means no embedding API costs passed to you, and no surprise charges when you import large document sets. LadybugDB (our active fork of Kùzu) handles all graph operations in-process.
Can I pay with FET or ASI tokens?
Yes — pay with FET/ASI tokens via the Fetch.ai billing system for a 15% discount vs. credit card pricing.
Is there a free trial for paid tiers?
Pro and Builder come with a 14-day free trial. No credit card required to start.
Do you offer annual pricing?
Yes — both Builder and Pro are available at 10× monthly price (effectively 2 months free) when billed annually.
Can I self-host instead?
Yes — GraphPalace ships a Docker Compose stack (Dockerfile.standalone) and WASM option. Run `docker compose up` to self-host. The managed service is easier; self-hosting is cheaper at scale. Usage metering is available via GET /v1/usage.
How does Agentverse Memory compare to Mem0?
Mem0 is a solid product. But they charge $249/month to unlock graph memory — leaving developers on the $19/month tier with vector search only. We include full knowledge graph support at every tier, including free. We also don't make LLM calls at write time, which means your memory costs don't scale unpredictably with ingestion volume.
How does Agentverse Memory compare to RetainDB?
RetainDB has made graph memory more accessible at $20/month — that's a real improvement over the market. But RetainDB uses LLM extraction at write time, which means 2–3 seconds of write latency per operation and token costs that scale with ingestion volume. Agentverse Memory uses TF-IDF + LadybugDB: no LLM at write time (0.035ms p95 in-process engine write op, measured in am-local), so writes carry no extraction latency or token bill, and graph is free from day one — not gated behind a $20/month tier.
Ready to build with graph memory?
Start on Explorer for free. Upgrade when you need more agents or ops. Cancel anytime.
All prices in USD. Annual billing saves ~17%. Credit card (Stripe) or FET/ASI token payment via Fetch.ai billing. SOC 2 report available on request.