Engine Pricing
Pricing preview. Mpalo is early-stage and pre-launch. The rates and plan volumes below are planning assumptions for launch and are not currently available for purchase.
Palo is not an LLM provider. Palo does not generate the final assistant response. It processes, stores, retrieves, traverses, and renders memory context for your own LLM.
The primary billable operation is Memory Encoding: transforming user input into a durable, experience-aware memory representation. Recall, rendering, traversal, mapping are separate Memory Operations based on how deeply your product uses the memory layer.
Architect includes 120M Memory Operations per month. SuperBloom includes 1B Memory Operations per month. Overage is billed at the operation rates below.
Prices cover Palo Memory Operations only. External LLM prompt and completion usage is billed separately through your own provider keys.
| Engine | Memory Encode | Recall + Render | Traversal | Mapping | Best For |
|---|---|---|---|---|---|
| Palo Nano mpalo-palo-nano | $0.235 | $0.355 | $0.18 | $0.24 | High-volume, latency-sensitive personalization |
| Palo Bloom mpalo-palo-bloom | $0.806 | $1.032 | $0.54 | $0.76 | Production agents and copilots with durable episodic memory |
| Palo Research mpalo-palo-research | $1.80 | $2.34 | $1.26 | $1.68 | Long-horizon synthesis and complex temporal reasoning |
Rates are per 1M operation-specific memory tokens.
Memory Encode: creates Palo's latent memory embedding and rich, experience-aware memory representation.
Recall + Render: retrieves stored memories and renders them into context for your external LLM.
Traversal: follows a short episodic chain of related memories.
Mapping: explores multiple temporal memory paths for deeper reasoning.
Depth is optional: basic memory is affordable. Traversal and Mapping are enabled selectively when an application needs deeper retrieval.
Live Episodic Memory is for real-time interactions, agents, user events, conversations, and ongoing personalization.
Bulk Memory Ingest is for documents, archives, exported conversations, transcripts, and large pasted context. It keeps deeper retrieval modes off by default for cost-efficient memory creation at scale.
Monthly Cost Calculator
Recall + Render is enabled by default. Traversal and Mapping are off until you select them. External LLM usage is billed separately through your own provider.
Estimated Monthly Total
$0.00