Pearl

Sign in to add this project to your watchlist

Overview

Tickers
PEARL PRL
Sector
decentralized compute
Lifecycle stage
Mainnet
Pre-TGE classification
Post-TGE
Token status
launched
Last updated

Pearl Research Labs’ proof-of-useful-work layer 1 pairs matrix-multiplication mining with an AI inference business. Its native coin launched in April 2026. The useful share of mining and current integration economics remain unverified; fetched pages disagree on availability of its Together AI endpoint.

Token status summary

launched

LearnPearl dates Pearl’s public layer-1 mainnet to 27 April 2026; ChainCatcher’s October report independently refers to the token launching at the end of April. This places the native coin inside the rolling 12-month inclusion window as of 13 November 2026.

Source: LearnPearl editorial team — dated status analysis, not a mining or rewards guide

Synthesis

No synthesis has been written yet.

Product

Pearl publishes an OpenAI-client example pointing to its inference API. No authenticated inference request was made during this review.

Supported

Project announcement

Evidence (1): 1 supporting

Pearl advertises serverless inference, managed inference with dedicated capacity, and an on-premise serving stack. The offer is distinct from evidence of paying customers or measured performance.

Supported

Project announcement

Evidence (3): 3 supporting
  • supports: Pearl Research Labs
    Managed InferenceDedicated capacity for sustained workloads: reserved throughput, private routing, and custom or fine-tuned weights with better economics than the public endpoint.Book a Call
  • supports: Pearl Research Labs
    On PremiseOur serving stack deployed inside your own environment. Your hardware, your region, your controls, with inference traffic that never leaves your network.Book a Call
  • supports: Pearl Research Labs
    ServerlessOpenAI-compatible endpoints on our optimized serving stack. Pay per token, with no clusters to reserve, no idle GPUs, and nothing billing between requests.Get API Key

The INT whitepaper explicitly says mined tokens are not automatically convertible into useful compute and describes an on-chain compute marketplace as a future possibility.

Supported

Project announcement

Evidence (1): 1 supporting

Independent reproducible benchmarks for the current serving stack, service-level performance, and savings net of token subsidies were not established in the fetched sources.

Not yet verifiable

No evidence recorded.

Pearl describes a proof-of-useful-work layer 1 in which mining can be a by-product of matrix multiplication; this describes its design, not the share of mining serving paid AI demand.

Supported

Project announcement

Evidence (1): 1 supporting
  • supports: Pearl Research Labs on GitHub — master README
    Pearl is an L1 blockchain based on the **Proof-of-Useful-Work** protocol, where mining is done as a by-product of arbitrary matrix multiplication, as proposed [in this paper](https://arxiv.org/abs/2504.09971).

Pearl publishes per-token serverless prices and describes them as launch discounts; this is a price list rather than proof of realized savings.

Supported

Project announcement

Evidence (1): 1 supporting
  • supports: Pearl Research Labs
    Serverless inference pricing for Pearl Inference Platform models, now available at special launch discounts. Pay only for the tokens you use.
Back to sections

Team

Hashrate Index attributes a Princeton PhD, Columbia experience, a Hebrew University post and prior Nvidia/Vast Data work to Weinstein. Its author works in Luxor marketing, so this remains unverified here pending academic or employment records.

Unverified

Evidence (1): 1 supporting

Together AI names Omri Weinstein as Pearl Research Labs co-founder and CEO; that commercial partner is not independent confirmation of the full team or his background.

Unverified

Evidence (1): 1 supporting

The full executive and engineering roster, headcount, legal entity records and independently checked employment histories were not established. No personal or residential address was collected.

Not yet verifiable

No evidence recorded.

Back to sections

Funding

TheMarker original refused curl access. Signed financing terms, investor confirmations, valuation, financing dates, GPU-fund structure, deployment of proceeds and runway remain unverified; the two financing channels should not be treated as a single token sale.

Not yet verifiable

No evidence recorded.

ChainCatcher attributes to TheMarker an October report of $130 million: $55 million equity financing and $75 million for a GPU infrastructure fund, naming Lightspeed, Ribbit and Square Peg. This supports the existence of that attributed report, not independently verified closing, cash received or fund terms.

Supported

Evidence (1): 1 supporting
Back to sections

Technical progress

An independent protocol, proving-system or wallet audit report, its scope and remediation record were not established in the fetched material.

Not yet verifiable

No evidence recorded.

The INT whitepaper specifies a 194-second target block interval, a Bitcoin-derived UTXO design and a non-Turing-complete stack machine with OP_CAT. These are published design parameters, not a claim that all implementation paths were audited.

Supported

Project announcement

Evidence (1): 1 supporting

At retrieval the project explorer labeled the network Mainnet and listed block 142510 among its latest blocks. This is a project-operated snapshot, not independent node verification or a measure of AI demand.

Supported

Project announcement

Evidence (1): 1 supporting

The public repository README says the monorepo contains a full node, wallet, SPV light client, ZK proving system, vLLM miner and supporting tools. Repository publication does not establish production correctness.

Supported

Project announcement

Evidence (1): 1 supporting

The fetched release feed publishes v1.5.7 on 12 November 2026. Its notes include Prometheus metrics, dependency updates, a testnet rank-penalty fork reschedule and P2P security hardening. No binaries were installed or run.

Supported

Project announcement

Evidence (3): 3 supporting

A deployed FP8/V4 mainnet activation and compatibility with the current release were not independently established. The new release feed alone does not resolve this question.

Not yet verifiable

No evidence recorded.

The v1.5.5 release notes describe wrong-network address rejection, removal of partial wallet post-quantum support, and hardening of orphan handling, SPV sync and peer resource limits. Earlier post-quantum messaging should not be read as evidence of current wallet support.

Supported

Project announcement

Evidence (1): 1 supporting
Back to sections

Adoption

Paying inference customers, production requests, retained customers, revenue, utilization and the current fraction of mining associated with outside AI demand were not independently established. Blocks and transfer counts do not answer those questions.

Not yet verifiable

No evidence recorded.

The current availability of the Together serverless Pearl endpoint is disputed in the fetched pages: Pearl’s network page calls it live, while Together’s model page says it is unavailable. The original May announcement is historical and does not resolve the present contradiction.

Disputed

Evidence (2): 1 supporting, 1 contradicting; 1 needs review

Together AI announced an exclusive Pearl partnership and a Gemma-4-31B-it-pearl endpoint discounted by over 25%, funded by expected emissions, on 15 May. As a commercial partner, its announcement does not independently verify the subsidy mechanism, realized savings or current service availability.

Unverified

Evidence (1): 1 supporting
Back to sections

Competitors

Pearl’s INT whitepaper positions matrix-multiplication mining against Bitcoin’s hashing. This comparison concerns the work function and monetary design rather than an independently measured advantage.

Supported

Project announcement

Evidence (1): 1 supporting

A neutral comparison with other decentralized compute networks, centralized inference providers and alternative useful-work protocols was not established, including comparable prices, accuracy, demand and integration overhead.

Not yet verifiable

No evidence recorded.

Blockworks argues that serverless providers, which control both hardware and AI software, are the natural early adopters, unlike hosts whose tenants control the GPUs. A co-author discloses PRL holdings, so this is retained as interested analysis.

Unverified

Evidence (1): 1 supporting
Back to sections

Token status

The selected asset is a native layer-1 coin, not an established ERC-20 or SPL contract match. No wrapped, bridged, exchange or same-ticker contract was authenticated to an official deployment in this review; their relationship remains unknown.

Not yet verifiable

No evidence recorded.

LearnPearl dates Pearl’s public layer-1 mainnet to 27 April 2026; ChainCatcher’s October report independently refers to the token launching at the end of April. This places the native coin inside the rolling 12-month inclusion window as of 13 November 2026.

Supported

Evidence (2): 2 supporting

Pearl says the company and team will lock at least half their holdings for six months from 9 November 2026. The statement does not establish enforcement or apply that commitment to all directly held shareholder coins.

Supported

Project announcement

Evidence (1): 1 supporting

Wallet-level ownership, the holdings denominator against issued supply, sale history, lock addresses, lock enforcement and the holdings of other investment firms remain unverified. A share of eventual maximum supply is not a share of current circulation.

Not yet verifiable

No evidence recorded.

Pearl says it launched without a premine, opened mining to anyone and has not sold PEARL to any party. This is a project disclosure, not an independently reconciled genesis distribution.

Supported

Project announcement

Evidence (1): 1 supporting
  • supports: Pearl Research Labs (FAQ decoded from page hydration JSON)
    The Pearl network launched with no pre-mine. Mining was open to anyone from the start, and within days of launch Pearl Research and its associates made up a small fraction of the network’s mining activity. Pearl Research has not sold PEARL to any party, over the counter or otherwise.

Pearl’s current holdings disclosure also uses PEARL. Both PRL and PEARL are retained as project-used symbols; identical symbols on other chains are not automatically matches.

Supported

Project announcement

Evidence (1): 1 supporting
  • supports: Pearl Research Labs (FAQ decoded from page hydration JSON)
    As of November 9, 2026, to the best of our knowledge, Team and Shareholders hold approximately 98 million PEARL in aggregate, or about 4.6% of total supply. This consists of approximately 72 million PEARL held by Pearl Research on behalf of its team and shareholders, and approximately 26 million PEARL held directly by Pearl Research shareholders.

Pearl’s project network page uses PRL for mining emissions. PRL is stored as a source-backed symbol, not a unique project identifier.

Supported

Project announcement

Evidence (1): 1 supporting
  • supports: Pearl Research Labs (FAQ decoded from page hydration JSON)
    Gemma-4-31B-it-Pearl, an instruction-tuned checkpoint running on Pearl’s Proof of Useful Work, is live as a serverless endpoint on Together AI at 25%+ below the standard price, funded by the PRL mined on the same forward pass.

Enforceable governance, equity, revenue-share, redemption and compute-access rights were not independently established. In particular, company equity financing should not be presented as financing for, or rights accruing to, coin holders.

Not yet verifiable

No evidence recorded.

The reference subsidy code gives genesis no subsidy and uses a default emission constant of 650,226 blocks; this is a code-reading observation, not an independently validated circulating supply.

Supported

Project announcement

Evidence (1): 1 supporting

Pearl’s reference subsidy code sets maximum supply at 2.1 billion coins. The whitepaper describes smoothly declining issuance rather than discrete halvings; no independent node replay was performed.

Supported

Project announcement

Evidence (1): 1 supporting

Pearl reports approximately 98 million PEARL held by the company, team and shareholders on 9 November: roughly 72 million held by Pearl Research for team/shareholders and 26 million held directly by shareholders. It describes this as about 4.6% of total supply and says these coins were mined publicly or bought at market prices. This is self-reported, not a wallet reconciliation.

Supported

Project announcement

Evidence (1): 1 supporting
  • supports: Pearl Research Labs (FAQ decoded from page hydration JSON)
    As of November 9, 2026, to the best of our knowledge, Team and Shareholders hold approximately 98 million PEARL in aggregate, or about 4.6% of total supply. This consists of approximately 72 million PEARL held by Pearl Research on behalf of its team and shareholders, and approximately 26 million PEARL held directly by Pearl Research shareholders.
Back to sections

Risks

Current pool concentration, majority-attack cost, protocol-upgrade decision rights, release signing and operational control were not independently reconciled. Neither a public code repository nor a proof-system paper establishes deployed system security.

Not yet verifiable

No evidence recorded.

Blockworks estimates that serving AI while mining is currently less efficient than dedicated mining and frames FP8 as critical to integration economics. These are dated estimates from a report with a PRL-holding co-author, not independently accepted performance facts here.

Unverified

Evidence (1): 1 supporting

LearnPearl states that Pearl Research Labs does not fund, review or approve its work and that it currently has no commercial partners, sponsored links or referrals. This publisher disclosure supports treating its status analysis as independent editorial work; it is not an external audit of staff holdings.

Supported

Evidence (1): 1 supporting

The INT whitepaper frames native settlement of compute contracts as a future possibility. Coin issuance alone should not be interpreted as a claim on inference-company revenue or a currently enforceable compute entitlement.

Supported

Project announcement

Evidence (1): 1 supporting

Jurisdiction-specific legal treatment, service liabilities, financial disclosures and regulatory compliance were not established; no conclusion about compliance or noncompliance is drawn.

Not yet verifiable

No evidence recorded.

Dependence on particular hardware, inference providers, rented capacity and the economic sustainability of subsidy-funded pricing have not been measured independently for the current deployment.

Not yet verifiable

No evidence recorded.

Pearl’s own disclosure establishes that its no-premine narrative does not mean the company, team and shareholders own no coins: it reports aggregate holdings and distinguishes them from unrelated investment firms that may have business relationships with Pearl. Independent ownership concentration remains unknown.

Supported

Project announcement

Evidence (1): 1 supporting
  • supports: Pearl Research Labs (FAQ decoded from page hydration JSON)
    Other investment firms and individuals, some with significant holdings, have acquired PEARL on their own through mining or open-market purchases. Pearl Research was not a party to those acquisitions, and these holders are not Pearl Research equity holders. Pearl Research may have, or may develop, business relationships with some of them. Pearl Research welcomes other holders and interested parties to reach out to discuss how they can best support the network’s growth and adoption.
Back to sections

Sources

Back to sections
Back to discovery

Nothing on this site is financial advice.