Recording: https://youtu.be/fimDRdL1Zsk
Host: Brandon Kramer (BK) from The Graph Foundation
Rewards Eligibility Oracle: Two-Month Retrospective
The Graph Foundation and Samuel Metcalfe reviewed REO from August 25 to October 29, 2026.
Why REO exists. In 2025, 15.2% of indexing rewards went to Indexers that were not serving queries or indexing data that consumers used. REO pays indexing rewards only to Indexers that serve queries. Total issuance does not change; REO changes who receives it.
How eligibility works:
- An Indexer needs at least 5 online days in a rolling 28-day window.
- A qualifying query returns HTTP 200, responds in under 5,000 ms, and is within 50,000 blocks of chainhead.
- The oracle runs daily at 10:00 UTC and renews eligible Indexers on-chain for 14 days.
Three steps: Each with 14 days of notice
- Step 1 (from August 25): 1 qualifying Subgraph per online day. Eligible Indexers went from 51 to 55, and none lost eligibility during the step. Reclaims were 9.8% of indexing issuance.
- Step 2 (October 6 to 26): 5 qualifying Subgraphs per online day. Eligible Indexers went from 55 to 48. Rewards per day held steady: 636K GRT a day, against 622K GRT in Step 1, shared by 47 Indexers instead of 52.
- Step 3 (from October 27): 10 qualifying Subgraphs per online day. This step is still settling. 39 and 38 Indexers served 10+ Subgraphs on October 27 and 28. Renewals made before October 27 run their full 14 days, so the full effect shows by about November 10.
About 667K GRT from failing Indexers went to the reclaim contract over the period.
Delegators: check where your stake sits. On October 29, 22 of 69 active Indexers were failing REO, and 79.0M GRT was still delegated to them. Those delegations earn no indexing rewards. Find your Indexer on the REO dashboard and check its status and online days. Undelegating starts a 28-day thaw. Check out the dashboard here: https://reo.thegraph.com/
REO Strategy Shift: Phase 2 (Step 4)
Indexers above 40M GRT total stake: Work out your Step 4 number before November 17.
The shift. Steps 1 to 3 raised the baseline for every Indexer (1, then 5, then 10 qualifying Subgraphs), so every eligible Indexer now serves queries on at least 10 Subgraphs. From Step 4, the focus moves from the number of Indexers to the amount of stake. Indexing rewards depend on stake, and stake is concentrated: the 10 largest Indexers hold about 70% of total stake and received about 71% of indexing rewards (stake as of November 2, 2026; rewards paid August 2 to October 29, 2026).
- Phase 1 (Steps 1 to 3): the same baseline bar for every Indexer. Rewards stop going to Indexers that do no measurable work.
- Phase 2 (from Step 4): the bar scales with total stake. The Indexers that receive the most rewards do the most work.
Step 4, from November 17. Required qualifying Subgraphs per online day = the greater of 10, or total stake divided by 4M GRT, rounded up.
| Total stake | Today | Step 4 | End goal |
|---|---|---|---|
| 30M GRT | 10 | 10 | 30 |
| 50M GRT | 10 | 13 | 50 |
| 100M GRT | 10 | 25 | 100 |
| 200M GRT | 10 | 50 | 200 |
Only Indexers above 40M GRT total stake see a change at Step 4. Everyone else keeps the 10-Subgraph minimum.
- End goal: 1 qualifying Subgraph per 1M GRT of total stake. Each later step gets at least 14 days’ notice and the data behind it.
- Serve, not sync: the count is Subgraphs you serve qualifying queries on. Plan sync capacity with headroom.
- Before November 17: count the qualifying Subgraphs you serve each day. If you are short, allocate to and sync more.
Dashboard: https://reo.thegraph.com/
Criteria: rewards-eligibility-oracle/ELIGIBILITY_CRITERIA.md at main · graphprotocol/rewards-eligibility-oracle · GitHub
Ongoing Foundation initiatives
- Multi-Gateway Plan: an internal Foundation proposal before it goes to The Graph Council for review. Deep dive below.
- Liquid Staking: holding in Phase 1, a small soft launch. Phase 2 brings UI updates and a new liquid staking partner. Informational only, not financial advice.
- Direct Indexer Payments (DIPs): the contracts are deployed on-chain but dormant, and they do not change indexing rewards today. Release is slated for later this year.
- Subgraph Studio pricing: an internal Foundation proposal, in progress before any changes reach users.
Multi-Gateway Plan
BK and Pedro walked through the community version (v0.2) of the Foundation’s Multi-Gateway Strategy: how the Foundation runs its canonical gateway and supports external Subgraph gateways.
What a Subgraph gateway does. A gateway business routes each query to the best available Indexers, bills consumers and pays Indexers through Graph Tally, issues API keys, and manages Subgraph versions and publishing. Some also run their own indexing or curation.
Four Foundation commitments
- Run the canonical gateway: long-term, run by the Foundation, and run for profit. It stays the default in Studio, on thegraph.com, and in the Docs.
- Enable external gateways: three tiers of support (Enablement, Endorsement, and Expansion). Only Enablement is open to any team by default.
- Limited influence by default: at Enablement, the Foundation has no say over a gateway’s product, pricing, or marketing, and no access to its IP, users, or accounting.
- Not prioritizing outreach: the Foundation is not actively pursuing new gateways, but continues ongoing conversations with design partners.
Why the Foundation runs a canonical gateway
- A canonical product offering: the gateway, Studio, and Explorer let developers build, publish, pay, and query with no infrastructure.
- Limiting outside leverage over the protocol: no single outside provider can set terms for everyone.
- Dogfooding: real query volume shows where the network falls short, and fixes go back into the open-source gateway.
Ground rules for the canonical gateway. It runs as a profitable business, publishes QoS data to the same standard it asks of others, and uses the same open-source software and payment rails as any operator. It will not use a loss-leader model or price Subgraph queries below cost, and it gets no privileged routing.
Three tiers of external support
- Enablement (open to any team): open-source Graph Node and gateway components, setup help until launch, and historical context. The support budget is 2 to 4 hours per week across all Foundation staff. Enablement brings no public mention or endorsement.
- Endorsement: the gateway commits to a QoS threshold, public QoS data, and brand rules. The Council approves the criteria, not each endorsement, and gets two weeks’ notice of each endorsement. Endorsed gateways get acknowledgement on X, LinkedIn, Discord, and the Forum.
- Expansion: paid access to product, marketing, BD, and event reach through a negotiated contract. It requires Endorsement first and includes auditing, reporting, and disclosure of paid placement.
Scope. The proposal covers Subgraphs only. Substreams and future data services each get their own strategy when they need one.
Next steps
- Working with two potential design partners
- Developing a QoS threshold for Subgraph gateways
- Developing brand and referral guidelines
- Developing a multi-service gateway framework
- Updating the Docs to be provider neutral
- Updating the Docs for the collaboration framework
Additional Updates
Chain integrations
The chain deprecation schedule is complete. botanix and botanix-testnet were removed on November 2, the last deprecation on the schedule.
In total, 14 chains are now deprecated and removed from Subgraph Studio: botanix, botanix-testnet, near-mainnet, near-testnet, fuse, fuse-testnet, soneium, soneium-testnet, kaia, kaia-testnet, zksync-era-sepolia, arbitrum-nova, monad-testnet, and viction. Soneium keeps its Substreams support.
ENSv2 Subgraph
Indexers: start syncing the ENS v2.1.1 Subgraph now. It went live yesterday, November 2, and a large share of ENS queries will move to it. ENS has not yet confirmed whether v2.1.1 is the final version.
- October 27: ENSv2 went live on Ethereum mainnet, and ENS v2.0.1 went live the same day.
- November 2: the ENS v2.1.1 Subgraph went live.
Adjacent ecosystems
Changes in other ecosystems that affect Subgraph and Substreams data. Dates come from project docs and news coverage and can still slip.
Recent and scheduled
- ENSv2 (live October 27): the v1 registrar no longer takes new .eth names. Existing names keep resolving. See the ENSv2 section above.
- BSC Jenner on testnet (October 28): the hard fork is live on the Chapel testnet. The mainnet date is not announced.
- Unichain to OP Enterprise (scheduled October 29): Optimism takes over production operations. The chain ID and addresses stay the same.
- Polymarket Protocol V2 (November 2, target): new markets move off the CTF contracts to a single positions contract with pUSD collateral. Existing positions are not force-migrated, so V1 and V2 run side by side.
Watch list
- Ethereum Glamsterdam: the Hoodi testnet fork is still postponed, with no new date. Glamsterdam on Hoodi Postponed | Ethereum Foundation Blog
- Base Denim: native 200ms blocks replace Flashblocks. Mainnet is targeted for November. Upgrades - Base Documentation
- BSC Jenner mainnet: brings Payment Lane and millisecond block timestamps to mainnet. No date is announced yet.
Questions or feedback? Reply below, or join the next call on Tuesday at 17:00 UTC on the Discord stage.