Teams building on Solana rarely run their own validators just to read state. They buy access to someone else's indexing stack, and the four names that come up most often are Helius, QuickNode, Alchemy and The Graph. The pricing pages look superficially comparable — a free tier, a $49 entry plan, a $499 mid tier — but the metering units underneath are different enough that comparing headline numbers side by side produces misleading conclusions. Here is what each vendor actually publishes as of this writing, and where the comparison breaks down.
Helius: Solana-only, with streaming tied to plan tier
Helius sells Solana infrastructure exclusively, and its plan structure reflects that focus. According to the Helius pricing page, the ladder runs: Free at $0 per month with 1M credits and 10 requests per second; Developer at $49 with 10M credits and 50 RPS; Business at $499 with 100M credits and 200 RPS; Professional at $999 with 200M credits and 500 RPS; and a custom Enterprise tier starting at 1B credits.
The credit allowance is the least interesting part of that table. What actually gates architecture decisions is which streaming and asset-indexing features unlock at which tier. Digital Asset Standard (DAS) API throughput — the endpoint most NFT and token-metadata applications depend on — is capped at 2 requests per second on Free, 10 on Developer, 50 on Business and 100 on Professional. LaserStream, the company's streaming product, is WebSocket-only on the free plan; gRPC access appears on Developer as devnet-only, and full mainnet gRPC arrives at Business. Staked connections, which matter for transaction landing rather than reads, also begin at Developer.
Two other details are worth flagging before committing. Data add-ons for high-volume streaming are listed as starting at $400 per month on the Professional plan, on top of the $999 base. And dedicated nodes are an Enterprise-tier item, not something a $999 plan includes.
QuickNode: bigger credit pools, multi-chain footprint
QuickNode covers many chains rather than specializing in Solana. Its published plans, per the QuickNode pricing page, are: Free Trial at $0 with 10M API credits and 15 RPS; Build at $49 per month with 80M credits and 50 RPS; Accelerate at $249 with 450M credits and 125 RPS; Scale at $499 with 950M credits and 250 RPS; Business at $999 with 2B credits and 500 RPS; plus custom Enterprise.
Annual commitments reduce the monthly figure — Build drops to $42, Accelerate to $212, Scale to $424 and Business to $849. Endpoint counts scale with the tier as well: one on Free Trial, 10 on Build, 20 on Accelerate, and 50 on Scale and Business. Support response targets are published as 24 hours on Build, 12 hours on Accelerate and eight hours on Scale and Business. SOC 1 and SOC 2 Type 2 certification is listed under Enterprise, which is the relevant line for anyone with a compliance review ahead of them.
Alchemy: usage-based rather than tiered
Alchemy takes a different shape. Rather than fixed monthly buckets, the Alchemy pricing page describes a free plan with 30M compute units per month and 500 CU/s of throughput — which the page equates to roughly 25 requests per second — followed by Pay as You Go at $0.45 per million CUs for the first 300M CUs and $0.40 per million above that, with 10,000 CU/s throughput. Enterprise is custom, with 1,000+ requests per second.
Alchemy publishes an average of approximately 27 compute units per request, which is the only vendor-supplied conversion factor in this group. Using that figure, the 30M CU free allowance works out to roughly 1.1M requests per month. That is the company's own stated basis for estimation, not an independent measurement.
The Graph serves Solana through Substreams, not a Hosted Service
This is where comparisons published elsewhere frequently go wrong, and it is worth stating plainly: The Graph's Hosted Service no longer exists. Its endpoints were retired on 12 June 2024 at the conclusion of the multi-phase "Sunrise" upgrade, and queries are now served by The Graph Network's independent indexers. Any guide still recommending that developers deploy Solana subgraphs to the Hosted Service is describing infrastructure that was decommissioned.
Solana support on The Graph arrived through Substreams — Rust modules that extract and transform chain data — in an announcement dated 3 November 2022. The Graph's own post framed this as "the first step in bringing subgraphs to Solana," which is a meaningful caveat: Substreams and subgraphs are not the same product, and a team expecting drop-in GraphQL subgraph parity with the EVM experience should validate that assumption before planning around it. The post claims "sync speed increases of over 100x with substreams parallelization." That is a vendor performance claim about historical indexing, published by the vendor, and should be treated as such rather than as a benchmarked result.
Why the credit numbers do not compare
At $49 per month, Helius lists 10M credits and QuickNode lists 80M credits. Reading that as an eight-to-one advantage would be a mistake. Each provider weights method calls differently based on the compute, memory and network cost of the underlying operation, so a credit at one vendor is not a credit at another, and neither is a compute unit at Alchemy. A workload dominated by heavy calls — large account scans, archival lookups, DAS queries with deep pagination — will burn through nominally generous allowances at rates that no cross-vendor table can predict.
The only reliable method is to take a representative sample of your actual method mix and price it against each provider's published per-method weights. Anyone quoting a single normalized "cost per million requests" across these four vendors is smoothing over the variable that matters most.
Picking a starting point
- Solana-only, needs DAS and streaming: Helius, with the caveat that mainnet LaserStream gRPC starts at the $499 Business tier.
- Multiple chains, wants headroom: QuickNode, where the $49 Build plan carries 80M credits and 10 endpoints.
- Unpredictable or spiky volume: Alchemy's Pay as You Go removes the tier-boundary cliff, and its 30M CU free allowance is the largest published here.
- Analytical or historical indexing: The Graph's Substreams, evaluated on its current decentralized-network form.
Pricing on all four changes without much notice. Confirm current figures on the vendor pages before you build a budget on them.