Helius Open-Sources rpc-perf-dash, a Global Solana RPC Benchmark With Anti-Gaming Safeguards

Helius hSOL$102.09+6.9% released rpc-perf-dash on July 22, an open-source system that runs continuous benchmarks against public Solana RPC providers across six global regions. Helius engineer Nick Pennie announced the tool on X that day, calling out existing benchmarks as "horribly flawed" for being non-global, running over home internet connections, ignoring tail latencies, omitting freshness and correctness checks, and cherry-picking time ranges.

The repository is licensed under Apache-2.0 and accepts community contributions. A live dashboard at helius.dev/benchmarks displays rolling results; the system collected more than 35 million samples in the 24 hours preceding publication.

Why Existing Solana RPC Benchmarks Fall Short

The core problem with provider-published benchmarks is that providers can optimize their services specifically for known test inputs, producing numbers that do not reflect real application behavior. When benchmark parameters are predictable, results can be engineered rather than earned.

rpc-perf-dash uses a commit-reveal scheme to address this. Before each challenge round, the system draws its parameters from live Solana chain state and commits to them under a hidden seed. That seed stays concealed until the challenge's time-to-live expires, so providers receive no advance view of the test inputs. After each round, anyone can verify that the parameters were fixed before responses arrived. Correctness is established by majority-vote consensus across all provider responses for each challenge.

Scoring formulas, consensus rules, and eligibility thresholds are documented in docs/methodology.md within the repository.

45 Methods Tested Across Six Regions and Four Cloud Providers

Per the repository, the benchmark tests approximately 45 RPC methods covering block and transaction lookups, account and token reads, and chain-metadata calls. Challenges run from six regions across four cloud providers (TSW, AWS, GCP, and Cloudflare), spanning US, EU, and APAC zones.

Per the benchmark's default scoring, each provider is graded on five weighted dimensions: latency at p50 and p95 (25%), win rate measured as the share of rounds where a provider returned the fastest correct response (25%), cross-region reliability (25%), data correctness (20%), and data freshness (5%). The weights are user-configurable, so teams prioritizing latency for trading applications can score differently from teams where correctness takes precedence.

Freshness is scored separately from correctness. A response can be accurate but stale, which rpc-perf-dash treats as a distinct failure mode relevant to any application reading real-time chain state.

Current Rankings on the Benchmark Dashboard

As of July 24, the dashboard ranked four providers against the default scoring weights. Helius led with 91.2 out of 100, ahead of QuickNode at 85.1, Triton at 78.6, and Alchemy at 70.4, per the dashboard.

Helius built rpc-perf-dash and currently ranks first in it. The open-source design and the self-run option are the primary checks on that conflict: the repository is public, the methodology is documented, and anyone can clone the project and produce independent measurements.

How to Add Providers and Verify Results Independently

The system is designed to add providers beyond the four it currently tracks; the README notes that additional providers can be included via configuration. Nathan Liow, who contributed to the project alongside Pennie, wrote on X that the goal is for the tool to "keep rpc providers accountable and accelerate the Solana ecosystem."

Helius has been expanding its developer tooling in 2026. In June, the team launched the balance-at API, which lets developers query any Solana wallet's historical SOL or token balance at a past timestamp in a single call.