Community Council Term 2: Six-Month Mid-Term Report
Period covered: April to September 2026. Figures current as of 7 September 2026 unless stated. Budget figures as of 27 August 2026.
1. State of play
Community Council Term 2 picks up where Term 1 left off. The term leans into community incentivisation programs and the reach of the Pyth brand. The council has also been developing ways to push the Pyth product into more B2C form factors to help demonstrate its usefulness and market positioning to consumers. This is most visible in PythFeeds.com, the Pyth Wheel and PythScan, all covered below.
Term 2 runs on an 8,000,000 PYTH twelve-month budget, with half paid in Q1.
Two of the three community products in this report came out of the March Community Hackathon: PythFeeds (Samurai), and Market DVR (Wattson), which has since developed into PythScan.
2. Content rewards system
The community deliberately shifted from a catch-all scattergun strategy to a much more defined, heavily community-first approach. As a result, engagement on X and beyond has trended up over the past six months.
Why the ‘Missions’ engine was retired
The term started on the broadcast Missions engine and ended in July. These are the numbers which forced the change:
| Month | Posts | Authors | Impressions | PYTH paid |
|---|---|---|---|---|
| April | 465 | 170 | 730K | ~330,000 |
| May | 5,811 | 1,969 | 4.75M | ~330,000 |
| June | 6,143 | 2,395 | 4.05M | ~330,000 |
Impressions per post fell from 1,570 (April) to 817 (May) to 659 (June). June added 22% more authors for 15% fewer impressions. Marginal participants added volume rather than reach.
The relaunch, 6 July
The creator system was relaunched as one path:
- Post in
#content-hub, or have the PythPulse tool find and surface your tweet. - Earn Impact Awards, administered by community leaders. Leaders are given a monthly allocation which is given away in full, with publicly auditable reasons.
- Get hand-picked as an Aristophanes: monthly PYTH stipend, 1.8x Impact Award multiplier, free Pyth Wheel spins, two-month trial then tenure.
- Publish in
#impact-ops, which everyone can read and only Aristophanes and community leaders can post to.
Community creators and leaders receive their own stipends, as well as multipliers on the Impact Awards they receive. Selection is based on content quality and track record ahead of follower count. The budget cap is roughly 15 contracted creators.
What changed in the output
Core community of 197 tracked handles, Pyth-related posts on X, retweets excluded.
| Month | Community impressions |
|---|---|
| March | 368K |
| April | 354K |
| May | 373K |
| June | 327K |
| July | 370K |
| August | 605K |
August is 69% above the March to July average, with zero ‘catch-all’ missions running.
- March to August total: 2.40M impressions, 104.6K engagement (4.4% engagement rate), 42 core members active per month on average, 27 active every single month.
- 1 August to 7 September: 46 core members active, 1,754 Pyth-related items, 695K views, 24.1K engagement.
- Impact Awards: July delivered 316 awards / 13,530 coins, against June’s 45 / 1,529, and effectively dormant in April and May.
The trade-off
The “everyone” contributor program peaked at 4.1M views in May and fell under 1M once the program was turned off. Core output went the other way.
This was a very deliberate tradeoff and the numbers prove why.
Engagement was up, but the quality was down significantly. The influx of bots on X through this period was an absolute scourge: automated and AI-generated replies inflated every metric while adding nothing, and they actively deterred genuine readers, who stopped engaging (rightfully) with content they deemed to be AI Slop. Broadcast incentives incentivised the behaviour. By June, 2,395 authors were producing 659 impressions per post, against 1,570 in April, on an openly exploited sybil surface.
Engagement from core community members, meanwhile, has grown throughout. The contracted creator system produces fewer impressions from identified people at a 4.4% engagement rate, inside a system where every award is published with a reason and every recipient is known.
Headline reach is down. Reach that survives contact with a sybil filter is up, and it is now produced by 46 people we can name.
Content Program budget: 1,090,670 of 3,000,000 PYTH used (36.4%). This is where the term’s remaining headroom sits.
3. Community stats
- Discord: 70,002 members. New joiners still in server: April 357, May 605, June 573, July 543. 1,165 unique members posted in the trailing 30 days to 3 August, across 60,977 messages.
- X, core community, March to August: 2.40M impressions, 104.6K engagement.
- PythPulse, the live mention scanner feeding the community Discord, launched 17 July. 2,123 Pyth mentions scanned in the first fortnight, 891 surfaced. An August upgrade added all-language coverage with inline English translation.
4. Pythenians 2.0
Governance thread: Pythenians II. Scope of work & budget request
Three forum stages, all under Community Council management, all led by Noname:
7 May, Pythenians 2.0 proposal (44 posts, 1,397 views, 100 likes). Treasury moved under a 6-of-7 Council multisig: 212 SOL staked (115 Helius, 97 Jito), 185 SOL liquid, ~$71,700 USDC, 69,000 PYTH. Collection 4,669 total, 3,682 minted. Pythenians Week discontinued in favour of Pyth Wheel, Pythentity and missions.
2 June, Key Proposal summary (15 posts, 485 views). Mint closed permanently at 3,682. Burn-to-level runs until the collection reaches 1,750, then stops forever. Three levels: Level 2 = hold one plus burn one; Level 3 = two more burned. Burn fees paid in PYTH, amount TBA. Pythentity badge system. Visual upgrade per level.
30 July, scope of work and budget request (11 posts, 426 views, 21 likes). One-month build. Team: Noname (project manager), Samurai (full stack), Kirito (graphic design), Sunshine (original collection artist). Budget: $2,000 from the Pythenians treasury, $500 per person. Deliverables: mint closure and collection authority under Council control, automated metadata and art regeneration on every level-up, handcrafted Level 2 and Level 3 art, and the PytheniansHub.
For context on cost: a full collection overhaul, a custom burn-and-upgrade dApp, and new artwork across two tiers, delivered for $2,000 total, by community members.
Status: community response unanimous in support.
Open question from the thread (31 August): who maintains the Hub after month one. The Council will run a nomination and procurement process to find community participants willing to operate the PytheniansHub. Anyone interested in getting involved is welcome to reach out or reply below.
5. Governance proposals
5.1 Community custom indices on Pyth Pro
Thread: Idea Proposal: Community Custom Indices on Pyth Pro — SCP, 16 August; 10 posts, 164 views, 20 likes
The idea. Let creators define baskets of existing Pyth feeds with weights and methodology, stake PYTH to create and keep an index live (10,000 to 50,000 PYTH suggested), and publish them as labelled “Community Index” feeds on Pyth Pro for venues such as Hyperliquid and TradeXYZ. Revenue share suggested at 40 to 60% creator, 20 to 30% DAO. Preferred technical path is Pyth Pro computing the index natively; the fallback is a third-party backend. Stated blocker: community members cannot currently publish feeds on Pro.
Feedback from the thread. Slashing and a higher stake plus a verification layer by Council, DAO or Douro (Mersault); a formal wind-down process so unstaking cannot force-close open positions, a recurring burned maintenance fee instead of one lock, no manual rebalancing after launch, and a smaller creator split (Lowkeigh); an interim upvotable “index hypothesis” campaign so the community learns index construction before any staking regime (Arguer). SCP’s synthesis on 22 August: permissionless creation, permissioned verification and listing.
Status: Ideas stage, no PIP drafted.
5.2 On-chain burn for Entropy protocol fees
Thread: Complementary Proposal to Pyth Reserve Evolution: Onchain Burn for Pyth Entropy Protocol Fees — SCP, 7 June; 8 posts, 206 views, 18 likes
The idea. On every Entropy request, burn 50% of the protocol fee on-chain and leave 50% withdrawable to the DAO, as a usage-equals-deflation complement to the Pyth Reserve proposal.
Where the community sits. Entropy fees are collected in native tokens (ETH, MON, HYPE), so an on-chain burn deflates those chains rather than PYTH. Marc confirmed no Entropy fees have yet been withdrawn from any chain, that withdrawal needs a Pythian Council contract upgrade, and that bridging native tokens to Solana is impractical. His counter: a single counterparty receives the accumulated native tokens across chains and delivers USDC to the DAO treasury, which folds into the existing monthly PYTH purchases. SCP accepted that direction on 10 July. The principle stands, in that Entropy revenue should create sustained PYTH buy pressure, with the implementation via counterparty rather than per-chain burns.
Status: idea stage, no vote.
5.3 Pyth Wheel Burn Registry
Thread: [OP-PIP] Implement PythWheel Burn Registry: 100 PYTH Permanent Burn per Spin Community Deflation Mechanism — SkyZ, 19 May; 18 posts, 475 views, 27 likes
100 PYTH permanently burned per spin, with spins allocated by community role. Derrp’s guardrails: one batched burn per month rather than per spin, a 50% cap on reserve tokens burnable at once, and a 12-month review. Marc moved it back to the Ideas Bank pending the funding-source question: prize pool, new budget line, or a separate burn fund.
Status: open as of 25 June.
6. Pyth Wheel
Overview
Pyth Wheel is primarily a tool to showcase Pyth Entropy in use across chains and in different contexts. It is also an accumulation mechanic for Pythenians holders, who receive free spins each week and accumulate PYTH on a weekly basis. This is one of several small utility mechanics attached to the NFTs that the community keeps as open secrets for the wider ecosystem to discover.
Usage and payouts
| Month | PYTH paid | Wallets | Spins |
|---|---|---|---|
| March | 41,635 | 356 | 2,254 |
| April | 37,686 | 378 | 4,646 |
| May | 44,163 | 439 | 5,469 |
| June | 40,411 | 416 | 5,009 |
| July | 100,000 | 431 | 5,600 |
| August | 93,958 | 497 | 7,001 |
| Total | 357,853 | 29,979 |
Wallet participation grew 40% from March to August. Spins per wallet went from 6.3 to 14.1, so the same audience is engaging more than twice as often. Cost works out at roughly 12 PYTH per spin.
The budget moved from roughly 40,000 to 100,000 PYTH per month from July as the Wheel graduated from Miscellaneous to its own line. The Wheel takes no fees; every contract address is public and payouts are verifiable per wallet at month end. Aristophanes receive free spins as part of the creator system, and Pythenians NFT holders receive free spins each week.
Which chain
Thread: Should We Move the Pyth Wheel Back to Base or Monad?
The Wheel is EVM-native, built on Pyth Entropy, and deliberately cycled across chains as an on-chain education tool. It launched on Base in March, moved to Monad, and to HyperEVM from September.
First week of September on HyperEVM (snapshot 6 September, at $0.055 PYTH and $0.11 per spin): 394 unique wallets and 2,941 spins, the strongest opening week of any month.
| Users in profit | 361 |
| Break-even | 2 |
| In loss | 31 |
| Median profit per user | $1.06 |
| Highest individual loss | $0.33 |
Every wallet is net positive over the lifetime of the Wheel across all three chains. A prize-weighting upgrade shipped alongside the migration.
The 2 September forum thread asking to move back to Base or Monad centres on spin cost, roughly $0.15 in HYPE plus bridging. The Entropy protocol fee on HyperEVM is unchanged at 0.0004 HYPE, so the difference comes from chain gas rather than Entropy.
Decision: stay on HyperEVM through September. Any net-negative wallets are equalised at month close.
7. PythFeeds.com
Overview
PythFeeds.com is a community-built retail market terminal on Pyth data, built and run independently by Samurai. The Council’s role is product direction, brand alignment and distribution. It is not an official Pyth product. All usage figures are Google Analytics 4, tracking live from 1 April 2026, snapshot taken 7 September 2026.
Usage
2,287 unique users and 3,197 sessions from 1 April to 6 September. Roughly half of those users arrived in the last five weeks.
Rolling 28-day active users: mid-May peak ~410, end-June trough ~180, late July ~450, 6 September ~1,050.
The last 30 days (8 August to 6 September) are the strongest on record. 28-day active users rose 132% on the previous 30 days. Weekly active users have held between 190 and 255 for five consecutive weeks; the previous best week was ~160 in mid-May.
Engagement:
| Metric | Value |
|---|---|
| Average session duration since April | 3m 43s |
| Last 30 days | 2m 06s (−56%) |
| Sessions per user since April | 1.4 |
New traffic is shallower than the early user base, which is normal when acquisition outpaces retention. Most visitors come once: retention is the weak point, ahead of acquisition, and it is the explicit priority for the next six months.
The read: the May spike lines up with the Missions program (May and June were the two heaviest mission months), and the June trough with missions stopping. The August to September surge happened with zero missions running, making it the first stretch of organic growth. Attribution is most likely organic: @PythFeeds posting on X, plus search results.
Note on figures. The GA4 “30-day active users” dashboard totals (56,587 for April to September) sum a rolling daily count and are not unique users. They should not be cited as monthly users. The “~10K monthly users” figure quoted when the workstream started is not supported by GA4; the peak trailing-28-day audience is ~1,050. We would rather post the smaller number that is true.
Development
- March: began at the Pyth Community Hackathon as Samurai’s entry, and grew from there.
- 28 May: workstream launched with Douro Labs and PDA. The direction from here is to use PythFeeds as a launchpad to trial a Pyth Pro consumer subscription package. Nothing has been finalised or formalised.
- 19 June: technical audit by Douro Labs. Findings: no product prioritisation, and an API layer proxying a single Pyth Pro key to PythFeeds users. The backend is Node.js; FMP is used only as a helper for PE ratios.
- 30 June: model changed to referral-only, arm’s length. PythFeeds stays as its own product and refers users to the Pyth Terminal for their own subscriptions and API keys. No redistribution of Pyth data. This removes the compliance-monitoring burden a formal partnership would have carried.
Consumer subscription trial
Nothing finalised or formalised. Pricing subject to approval and legal review.
It is being proposed that PythFeeds be used as a launchpad to trial a Pyth Pro consumer subscription package: the first attempt at a consumer-tier subscription on Pyth data, run on a community product rather than an official surface, where the community can test & iterate without committing the broader protocol to a model that has not been proven.
Working shape, subject to change: revenue split 70% to the DAO wallet, 30% to the builder, covering ongoing marketing, development and maintenance. Payments would run fully on-chain and verifiably, rather than through a conventional processor, keeping the product crypto-native and making the DAO’s share auditable by anyone rather than reported after the fact.
There is MUCH more to come here. If the trial works, the longer-term path is PythFeeds being folded into the Pyth Terminal as a community-built component, which would make it the first case of a community project graduating into the core product surface.
Account and content program
The @PythFeeds account on X is live, with a community content program running behind it. The intention is to transition this into a full-time position, funded from a share of the subscription revenue once the trial above is running, plus a monthly stipend from the Council’s community grants program. Part of the pay is tied to whether the product it markets actually earns.
8. PythScan
PythScan is a direct output of the March Community Hackathon run by the Community Council. Wattson (@MasterWattson) entered that event with Market DVR, built on the hackathon API the Council stood up, and went on to ship PythScan from it. It has been live for testing since 24 August on two surfaces.
What it answers
Plain-language questions about Pyth, restricted to official Pyth sources, with the bot built to say when it cannot verify something: live price, 24h / 7d / 30d and calendar-window change, price at a point in time, all-time high and low, two-asset comparisons, and top movers by market, region and period. It also covers the feed catalogue (does a feed exist, which feeds were added this week, feed and Hermes/Lazer ID lookups), governance (did a PIP pass, live Realms votes and those ending soon, council membership and elections, constitution and staking rules), and product questions (Core vs Pro, Pyth Pro pricing, recent updates, package versions, Pyth Pro revenue reports).
On X — @Pythscan
Tag it with a question and it replies through the filtered mention stream, with a rendered Pyth price, comparison, revenue or confidence card attached.
On Discord
Install it into a server and ask in plain language, in DMs, on mention, or in reply, plus six slash commands:
| Command | Does | Permission |
|---|---|---|
/setup |
pick reply channels and thread behaviour | Manage Server |
/alert |
price, new-feed and governance alerts, delivered by DM | any member |
/watchlist |
add, remove and list tracked assets | any member |
/updates |
automatic server posts for new feeds and governance | Manage Server |
/movers |
by direction, market, region and period | any member |
/brief |
today, yesterday, this week, last week | any member |
Build
Python 3.10 on discord.py and FastAPI, SQLite for state, and a Node renderer for card images. Roughly 76,700 lines of Python with a further 42,000 lines of tests across 124 test files, built by one community member in about two months.
Known limits
- Discord replies are truncated at 1,900 characters; X replies at 270.
- Alerts, watchlists and server broadcasts are Discord-only.
- It responds only when addressed (a mention, a DM, or a reply), and only to questions.
- Card rendering depends on a Node renderer that lives outside the repository.
Next: data visuals publishing to X.
Continuity note for the Council: the repository is currently private, single-contributor, without a licence or continuous integration. Securing a licence and a transfer path is a Cycle 2 deliverable, so a tool the ecosystem comes to rely on does not sit on one person’s account.
9. The PIRB
@thePIRB is a Pyth lore and bull-posting account run by Aristophanes from the creator program. It is the clearest example of what the contracted creator system produces that a mission brief never could: sustained narrative craft, built over months, with a visual identity of its own.
The past six months:
- Pyth lore and bull posting through AI video. The account’s core output is short AI-generated video, used to build a running mythology around Pyth rather than to restate announcements.
- Community creation and lore. It leans into the community’s own in-jokes, characters and running references, which is what gives the material something to attach to.
- Narrative craft first. The posts are weird, and deliberately so. They are built to stop a scroll and be remembered, which is a different job to informing an audience that already follows Pyth.
- A visual style that has compounded. The look has improved materially across the six months and continues to develop. It now reads as a recognisable identity rather than a series of one-off experiments.
The numbers, 19 March to 19 September. 719 posts, 124,751 impressions, 5,242 engagements at a 4.2% engagement rate. Three quarters of that volume is replies, which inherit whatever reach the thread they land in already had, so the figure that actually describes the account is its own content: 169 original and quote posts carrying 85,169 impressions at a 5.0% engagement rate.
| Month | Own posts | Video | Impressions | Mean / post | Median / post | Eng. rate |
|---|---|---|---|---|---|---|
| Mar (13 days) | 10 | 4 | 3,273 | 327 | 292 | 6.3% |
| April | 14 | 8 | 7,589 | 542 | 414 | 4.7% |
| May | 20 | 14 | 5,914 | 296 | 333 | 6.7% |
| June | 31 | 20 | 9,581 | 309 | 341 | 7.2% |
| July | 33 | 14 | 13,153 | 399 | 394 | 6.1% |
| August | 45 | 19 | 31,639 | 703 | 507 | 3.7% |
| Sept (19 days) | 16 | 11 | 14,020 | 876 | 660 | 4.7% |
Reach per post is up 2.7x since March on the mean and 2.3x on the median, while the engagement rate held in a 3.7% to 7.2% band rather than decaying as reach grew. Output rose alongside it: 31 to 45 original posts a month from June, against 10 to 20 before.
The distribution story. Total impressions are 872x the follower count; on own content alone, 596x. Follower count was flat across the window, so effectively none of this reach comes from the follower base. It comes from X distributing the videos to people who do not follow the account. Every one of the top ten posts reached between 8x and 63x the follower count on its own.
August is the inflection. That month the account worked out that short video tied to a live news hook or a recognisable meme format travels a long way past its own audience. The best-performing own post of the window, a video on the SEC and Hyperliquid with a Pyth angle posted the same day as the headline, took 7,591 impressions. Five of the six own posts that cleared 1,500 impressions are video, and all six are from August or September.
Discounting our own numbers. Roughly 20,000 of the 124,751 impressions, about 16%, came from five off-topic replies to large non-crypto accounts, which drew near-zero engagement. That is borrowed reach, and this report argues elsewhere that borrowed reach should not be counted. Discounted, the window total is ~104,750 impressions, and the honest own-content figure is the 85,169 above.
Source: X API v2, pulled 19 September 2026. Per-post data retained. Follower count at the start of the window is not recoverable through any available access path, so net follower growth cannot be stated.
10. RWA Markets research
rwa-markets.xyz is the work of Zinn (@zinnresearch), published under Refraction Research and sponsored and subsidised by the Community Council. It is the only public dataset of its kind: real-time analytics on tokenised real-world asset trading across decentralised and centralised venues, tracked at a granularity nobody else publishes.
Coverage at 9 September 2026:
| 24h volume tracked | $16.4B |
| Venues | 22 |
| Assets / markets | 536 / 2,218 |
| RWA share of tracked perp open interest | 18.6% ($11.6B) |
What it covers: nine dashboards across tokenised stocks, commodities, indices and treasury assets. Today (daily snapshot), Volume Dominance, OI Dominance, Top Traded Assets (with oracle data), Institutional Velocity (trade frequency, size, spreads, slippage), Asset Analyzer (per-asset liquidity and funding), Trader Intelligence (on-chain wallet analytics and capital flows), DeFi Markets (lending, yields, TVL, tokenised treasuries) and Spot Markets (tokenised stock and ETF spot trading and issuance).
Why it matters to Pyth. This is an important community run project that Pyth uses internally for data attribution on perpetuals volumes, alongside a range of other datapoints. Per Zinn, it fed the August RWA report. The protocol consumes it as working infrastructure.
Zinn has since built dedicated API connectivity specifically for Pyth, with partner access to the underlying database covering oracle provider attribution weighting per symbol, venue and wallet-level breakdowns, per-venue execution quality, and DeFi yield history. Pyth is the only party outside Refraction Research with this level of access. It is a working technical partnership between a community researcher and the protocol, built without a formal commercial agreement.
That is the clearest illustration of what the Council’s research budget buys. Zinn is producing work that is unavailable anywhere else, that the team relies on internally, and that carries the Pyth name into a market segment where nobody else has comparable visibility.
11. Budget
Figures as of 27 August 2026. All amounts in PYTH.
| Category | 6-month budget | Spent | % used |
|---|---|---|---|
| Role stipends | 280,000 | 306,486 | 109.5% |
| Content program | 3,000,000 | 1,090,670 | 36.4% |
| Hackathons | 300,000 | 133,000 | 44.3% |
| Council stipend | 420,000 | 420,000 | 100% |
| Miscellaneous (incl. Pyth Wheel) | n/a | 305,150 | n/a |
| Total | 4,000,000 | 2,255,307 | 56.4% |
Monthly spend against the recurring benchmark: April 588K, May 623K, June 405K, July 447K, August 121K, September 70K to date. May was the only month over benchmark, driven by hackathon payouts.
Role stipends is the only line over allocation, at 9.5% (26,486 PYTH) past its six-month envelope. The cause is growth rather than overrun: the role system expanded as more mechanics came online, and more community members wanted to dedicate serious time to running them. We would rather have that problem than the opposite one.
On the 43.6% unspent: this is deliberate, and it was planned for. The underspend sits almost entirely in the content program, because Missions were shut down mid-term and the replacement creator system only went live on 6 July. Rather than spend the difference to hit a number, we are holding it for the Hackathon, which returned more per PYTH than anything else the Council funded and deserves a materially larger prize pool and more of the Council’s time in the second half. The Cycle 2 increase we are asking for was already contemplated in the original request made at the start of the year.
What things actually cost
| Term 2 to date | |
|---|---|
| Pyth Wheel | ~12 PYTH per spin, ~720 PYTH per participating wallet |
| Hackathon | 133,000 PYTH for 40 submissions = 3,325 PYTH per submission |
| Hackathon, second iteration target | 500,000 PYTH for 400 submissions = 1,250 PYTH per submission, a 62% improvement in unit cost |
| PythFeeds, PythScan, SCPTech | a single stipend per project, capped at 20,000 PYTH per month and lower in several cases |
That last line deserves emphasis. PythFeeds, PythScan and the SCPTech work are each built by one person. There is no team, no agency and no vendor contract behind any of them. The entire development cost to the DAO is a monthly stipend to a single community member, capped at 20,000 PYTH and less in some cases. PythScan alone is 76,700 lines of Python with 42,000 lines of tests.
These are also long-horizon projects and should be judged as such. A retail terminal, an intelligence assistant and a governance tracking stack do not produce measurable returns inside a six-month window, and we would be misleading the DAO if we presented them as though they did. What they produce in Term 2 is working software, a user base to learn from, and optionality: PythFeeds is now the trial vehicle for a Pyth retail subscription, and PythScan is becoming the ecosystem’s front door for plain-language questions. The payoff sits in Terms 3 and 4.
