Introducing Paddock: Run Open AI Models on Your Own Hardware
Up to 37x faster than llama.cpp and faster than vLLM/SGLang on tuned models.
Read article →News, updates, and technical articles from the Truespar team.
Up to 37x faster than llama.cpp and faster than vLLM/SGLang on tuned models.
Read article →Torque 0.6.0 adds LLM-backed conversational and natural-language search, an MCP server for AI agents, curation, analytics, and crash-safe persistence.
Read article →Traverse 0.8 adds full ISO GQL alongside openCypher, a built-in APOC layer, vector and full-text search, and ML pipelines driven from Cypher.
Read article →Traverse 0.7 ships a graph data science library with 36+ algorithms callable from Cypher, and the full engine now runs in a browser tab via WebAssembly.
Read article →Torque 0.5.0 is an in-memory search engine in Rust: a drop-in Typesense replacement with 6x faster search, 7.5x faster ingestion, and GPU acceleration.
Read article →Traverse 0.6.0 adds performance gains, new Cypher features, and Silicon: a unikernel build that runs the whole database as a VM in under 100 ms.
Read article →Traverse is a new graph database, 7x faster than Neo4j and 3.6x faster than Memgraph. Built from scratch in Rust with full Cypher and Bolt support.
Read article →