TETRA vs Neo4j
You’re paying too much for Neo4j.
And it still OOMs on writes.
Same queries. Same hardware. Same Bolt protocol. We gave Neo4j twice the RAM budget. Neo4j used 10.8× more memory. Crashed on concurrent writes. Failed half the LDBC Interactive Complex queries. Costs $1,051/month for 16 GB of RAM.
TETRA ran the full suite in 66 MB. Passed all 14 LDBC queries. Stayed flat at 64 concurrent clients. $299/month flat.
Benchmark Results
We ran the benchmark. Here’s what happened.
105 Cypher queries. 30 iterations each. Both containerized. Neo4j got 1 GB RAM / 1 CPU; TETRA got 512 MB / 1 CPU. TETRA won 37, Neo4j won 19. TETRA completed all 11 writes; Neo4j OOM’d.
| Category | TETRA wins | Neo4j wins | TETRA speedup |
|---|---|---|---|
| Counts & Lookups | 5 | 0 | 1.1×–3.9× |
| Traversals | 5 | 2 | varied |
| Aggregation | 6 | 0 | 2.9×–63× |
| WITH Pipeline | 6 | 0 | 1.1×–22× |
| Real-World App | 7 | 0 | 1.4×–290× |
| Stress Test | 4 | 1 | varied |
| Total | 37 | 19 | |
| Write ops | 11/11 | 0/11 OOM | — |
Throughput under load
Neo4j drops from 90 q/s to 29 q/s at 8 clients. At 64, crashes. TETRA: flat at 485 q/s.
Pricing Comparison
Neo4j’s real cost.
Published pricing for Neo4j AuraDB tiers. TETRA is flat-rate — visualization, algorithms, and full openCypher included.
| Configuration | Monthly | Year 1 | Year 5 |
|---|---|---|---|
| AuraDB Pro 8 GB | $526 | $6,307 | $31,536 |
| AuraDB Pro 16 GB | $1,051 | $12,614 | $63,072 |
| AuraDB BC 8 GB | $1,168 | $14,016 | $70,080 |
| AuraDB BC 32 GB | $4,672 | $56,064 | $280,320 |
| TETRA | $299 | $3,588 | $17,940 |
Neo4j add-on costs (not included above)
- Bloom — $1,200–$2,500 / user / year
- Graph Data Science (GDS) — $10K–$25K+ / year
- Enterprise license — $20K–$200K+ / year
TETRA includes visualization, 30+ algorithms, and full openCypher at $299/month. No add-ons.
Migration
Migration in an afternoon.
100% Cypher TCK compliance — 1,611 / 1,611 scenarios passed. Same Bolt protocol. Point your existing Neo4j driver at TETRA’s Bolt endpoint.
Transparency
Where Neo4j wins.
19 of 105 queries. Primarily multi-variable RETURN projections and OPTIONAL MATCH chains. These are known optimization targets on our roadmap.
Ready to stop overpaying?
Same Cypher. Same Bolt. Less RAM. Lower cost. See the full technical breakdown or schedule a migration call.