Full product from day oneEvery plan gets every Synap capability. Tiers differ on volume and service level, not on what you can do.

Trial (Free, list $10/month — promotional offer, no card required): 12,500 credits/month, 1 agent / instance, hard cap, docs support, 7-day audit trail

Starter ($19/month launch offer, list $49): 62,500 credits/month, 3 agents / instances, email support (48h), $1.00/1K credits overage, 30-day audit trail

Pro ($249/month): 375,000 credits/month, 10 agents / instances, BYOK, email support (24h), $0.88/1K credits overage, 90-day audit trail

Scale ($999/month): 1,875,000 credits/month, 50 agents / instances, BYOK multi-provider, Slack support (4h SLA), $0.75/1K credits overage, 180-day audit trail

Enterprise (Custom): Volume-based credits, unlimited agents / instances, BYOK any provider, dedicated CSM, negotiated overage, custom audit trail

Every plan includes all retrieval modes (FAST/ACCURATE/HYBRID), full memory architecture & customization, agentic compaction, anticipatory retrieval, consolidation, A2A protocol support, REST & WebSocket APIs, webhooks & event streams, and custom memory schemas.At $19, Synap Starter offers full graph memory, all retrieval modes, custom architecture, anticipatory retrieval, agentic compaction, and GPT-4o extraction — capabilities that competitors at the entry tier (Mem0, Supermemory, Zep, Letta) either do not offer or gate behind their $249 plans.Synap pays for itself: Anticipatory retrieval and agentic compaction roughly halve the context tokens per turn, which on a typical bill is about 20% of total LLM spend; at $5,000 a month of spend that is about $1,000 a month back. Building the equivalent yourself is months of engineering and a standing maintenance cost to keep.Get started or read the docs. Talk to us about Enterprise.

Developer Pricing

Full product from day one.

Every plan gets every capability. Tiers differ on volume and service level, not on what you can do.

Every plan includes

All retrieval modes (FAST / ACCURATE / HYBRID) • Full memory architecture & customization • Agentic compaction • Anticipatory retrieval • Consolidation • A2A protocol support • REST & WebSocket APIs • Webhooks & event streams • Custom memory schemas

Promotional Offer

Trial

For indie developers

$10Free/month

12,500 credits / month
1 agent / instance
Hard cap (no overage)
Docs support
7-day audit trail

Start Free

No Card Required

Launch Offer

Starter

For projects going to production

$49$19/month

62,500 credits / month
3 agents / instances
Email support (48h)
Overage: $1.00 / 1K credits
30-day audit trail

Start Building

Pro

For growing startups

$249/month

375,000 credits / month
10 agents / instances
BYOK (Bring Your Own Key)
Email support (24h)
Overage: $0.88 / 1K credits
90-day audit trail

Go Pro

Scale

For scaled applications

$999/month

1,875,000 credits / month
50 agents / instances
BYOK (multi-provider)
Slack support (4h SLA)
Overage: $0.75 / 1K credits
180-day audit trail

Scale Up

Enterprise

For organizations at scale

Custom

Volume-based credits
Unlimited agents / instances
BYOK (any provider)
Dedicated CSM
Negotiated overage
Custom audit trail

Talk to Us

01Credits

How credits work

Every operation has a predictable cost. No surprise bills, no per-token math.

OperationCreditsWhat drives it

Retrieval, fast path1 per queryThe default inside a live turn

Retrieval, hybrid2 per query

Retrieval, accurate path3 per queryAdds decomposition and reranking

Cached readA fraction of oneMost in-conversation reads

Ingestion2 to 25 per documentLength

Compaction2 to 8 per callLength of the history being compacted

02Compare

Why Synap at $49$19?

Entry tier against entry tier. Here is how Synap Starter compares against the first paid plan of every major competitor.

Capability	Synap Starter $49$19	Mem0 Starter $19	Supermemory Pro $19	Zep Flex $125	Letta Pro ~$20
Graph memory	Full	Pro only ($249)	Semantic graph	Temporal graph	None
All retrieval modes	All	Basic only	Hybrid search	Graph search	Agent-controlled
Custom architecture	Full	Pro only ($249)	No	Custom types	Block schemas
Anticipatory retrieval	Yes	No	No	No	No
Agentic compaction	Full	No	No	No	MemGPT (limited)
Extraction model	GPT-4o	gpt-4.1-nano	Mini-class	Mini-class	Standard

At $19, you get what others charge $249 for. You just get less of it.

03ROI

Synap pays for itself

Anticipatory retrieval and agentic compaction roughly halve the context tokens per turn, which on a typical bill is about 20% of total LLM spend; at $5,000 a month of spend that is about $1,000 a month back. Building the equivalent yourself is months of engineering and a standing maintenance cost to keep.

Run the full comparison on your own numbers →

Evaluating alternatives?

See how Maximem Synap compares to Mem0, Zep, Letta, Supermemory, Cognee, and Evermind on verified LongMemEval and LoCoMo benchmarks, latency, and capabilities.

Compare Synap vs alternatives

04Billing

Billing questions

What does one credit buy?

A retrieval on the fast path costs one credit, a hybrid retrieval two, and an accurate-path retrieval three. Reads served from context already in your process cost a fraction of one. Ingestion costs 2 to 25 credits per document and compaction 2 to 8 per call, both driven by length.

What happens when I use my monthly credits?

On Trial, requests stop until the cycle resets; there is no overage. On Starter, Pro and Scale, overage is billed per 1,000 credits at the rate shown on your plan: $1.00, $0.88 and $0.75 respectively.

Is there an annual option?

Yes. Annual billing saves 17% on every paid plan.

What does BYOK mean here?

On Pro and above you can bring your own model-provider keys, so the models Synap uses for extraction and compaction run on your account and your terms. Scale allows more than one provider; Enterprise allows any.

05FAQ

Developer FAQ

How does Synap pricing scale?

Synap pricing is based on API calls and active user counts. Overage is charged at the per-unit rate published on every plan.

Can I self-host Synap?

Enterprise plans include an on-premise deployment option. Contact our team for architecture details and requirements.

What AI models does it work with?

Synap is model-agnostic. It works with OpenAI, Anthropic, Google, Mistral, Llama, and any LLM provider through our standard API.

Is there a free trial for paid plans?

Yes — the Growth plan includes a 14-day free trial with full feature access. No credit card required to start.

Still weighing this against building it in-house? Build vs buy agent memory walks the real cost, and measuring agent memory covers how to judge either option.

06Get started

Ready to build with memory?

Start with the free tier. Ship persistent memory in your AI app.

Book a Demo