Skip to main content

EVAL Engine

EVAL Engine (EVAL) — a reference card for a crypto project: what the product does and how verifiable the project’s key claims are; also — alignment with the Decentralized AI thesis (DecAI Fit).

EVAL Engine — an AI/Analytics | Infra project: a protocol/service for evaluating the quality of AI agents’ responses (including crypto‑native agents), recording the results in onchain storage (Chromia) for a verifiable “proof”.

Disclaimer

Important: this page is a reference overview of the project. Within DecAIHub, we separately record a verification based on open sources within the AI thesis (what is real AI/onchain/why a token, and which sources this is based on).
This is not financial advice and not a recommendation to buy/sell/hold any asset. We do not assess price, returns, or entry/exit timing.
Conclusions are based on publicly available materials found as of the verification date. The absence of confirmation does not imply the absence of the technology.

TL;DR (for readers who are not into AI)

  • What it is: an engine/protocol that evaluates AI agents’ responses and produces a “performance score”, recording the results onchain for verification.
  • Who it’s for: developers/operators of AI agents and teams that need verifiable metrics of response quality/reliability.
  • Why onchain/token: Tier‑1 litepaper: evaluation results are stored in Chromia (gas‑free relational blockchain) as an immutable public log; Tier‑1: evaluation reports are paid for with the $EVAL token (subscription model).
  • What you can actually touch (Tier‑1): Playground for evaluating responses (for example, “Evaluate Reply Tweet”).

How it works (very briefly)

  1. The user does: sends an evaluation request (evaluation request) for a specific agent response.
  2. The system does: runs an “LLM‑as‑a‑judge” evaluation on several criteria, aggregates it into a composite score (weighted average), adds engagement signals (per the litepaper) and forms a report.
  3. The user gets: an evaluation/score and the evaluation history (analytics/benchmarking).
  4. Where the blockchain is: the final evaluation results are stored onchain in Chromia for transparency/auditability (Tier‑1).
  5. Token role (if any): $EVAL is used to pay for evaluation reports under a subscription model; the litepaper also describes the $EVAL ↔ $CHR interaction for computational resources.

Quick links: AI thesis fit

Project passport

Field Value
Project name EVAL Engine
Also known as (aliases) Evaluation Validation Architecture; EVAL Engine
Logo (logo) —
Token symbol (ticker) EVAL
Category AI/Analytics | Infra
Primary chain (chain) Base
Type non-EVM
Network status mainnet
Consensus (consensus) PoS (Base; Ethereum L2)
Launch date (launch_date) —
Project status (status) active
Token (standard) ERC-20
Token contract 0xdd78523217390bb0d49c7601e7e54c36d71622f0
Token model (token_type) utility (subscription fees for evaluation reports)
Max supply 1 000 000 000 EVAL (max total supply, BaseScan)
Issuance fixed cap (max supply)
Tokenomics (tokenomics, brief) Litepaper: evaluation reports are paid for in $EVAL (subscription model).
Governance —
Admins/keys —
Audit —
Bug bounty —
Team (team) Johnson Lai; Prem Kumar (litepaper authors)
Funding (investors/funding) —
AI component (ai_component) LLM-as-a-judge agent answer evaluation
Use‑case (use_case) on-chain verification of AI agent quality
DecAI Fit — AI thesis fit 5/6 — high alignment with the Decentralized AI thesis
DecAI Fit component summary ai_score 3 / onchain_score 3 / token_score 1 / evidence_score 3
Last verified 2026-02-11

General project overview

EVAL Engine (Evaluation Validation Architecture) is positioned as a decentralized framework for evaluating AI agents with a focus on crypto‑native agents (CT/social agents). The litepaper separately emphasizes the problem of the absence of metrics and the risk of hallucinations without standardized evaluation, as well as the goal — to make evaluations verifiable and auditable by storing the results onchain (Chromia).

Per Tier‑1 sources, the project provides a “playground” for evaluation (for example, the quality of replies to tweets) and describes the architecture: multi‑model evaluation (LLM judges), score aggregation, storage of results in Chromia and the associated token layer for paying for evaluation reports.

Features/modules

  • Playground: evaluation of response quality (Evaluate Reply Tweet) and other scenarios.
  • Benchmarking / Analytics: evaluation history and comparison.
  • LLM‑powered judges: multi‑LLM evaluation by criteria (truthfulness/creativity/accuracy/engagement).
  • Onchain storage: recording of evaluation results in Chromia (gas‑free relational blockchain).

Typical scenarios

  • Scenario 1: a developer submits an agent response for evaluation → receives a score/report → sees the history in analytics.
  • Scenario 2: a team compares agent/model versions via benchmarking → receives a comparable result.

Getting started (wallet/network)

  • What you need: access to the web interface (playground); to pay for evaluation reports — $EVAL tokens (per the litepaper).
  • Network/chain: Base (token) and Chromia (storage of evaluation results, per the litepaper).
  • Fees: Base gas is incurred for onchain operations with the token; the litepaper claims gas‑free storage on Chromia for evaluation results.

Architecture/tech (non‑AI)

  • Components (per the litepaper): ingestion layer (API), evaluation engine (orchestration), storage/contract layer in Chromia, engagement module.
  • Participant roles: clients (agents/teams), evaluation engine (evaluation provider), onchain storage (Chromia).
  • Where logic runs: computation of the evaluation — offchain; recording of results and their immutability — onchain (Chromia), the token — onchain (Base).
  • Dependencies: Chromia (FT4), web/API layer; LLM inference providers (Hyperbolic/Groq are mentioned in the litepaper).

Token/economics (expanded, no forecasts)

  • Token utility: litepaper: each evaluation report is paid for in $EVAL (subscription model).
  • Protocol necessity drivers: the token layer is described as a payment mechanism for access to evaluations (evaluation service).
  • Who pays fees: users/clients, when they pay for services and/or perform onchain operations with the token.
  • Where fees go: not established per Tier‑1 (the fee distribution model is not described explicitly).
  • Issuance/burn: max total supply 1B EVAL (Tier‑1: BaseScan).
  • Security/incentive mechanisms: the litepaper describes “double‑signed transactions” for storing results (client + provider), but the onchain economics of incentives/slashing is not disclosed.
  • Distribution and token issuance schedule: not established per Tier‑1.
  • Treasury: not established per Tier‑1.

Governance

  • Who makes decisions: not established per Tier‑1.
  • How protocol parameters change: not established per Tier‑1.
  • Where decisions are discussed/made: not established per Tier‑1.
  • Keys/multisig: not established per Tier‑1.

Security and incidents

  • Audits: not found per Tier‑1.
  • Bug bounty: not found per Tier‑1.
  • Incidents: —.
  • Threat model: “LLM‑as‑a‑judge” risks (prompt injection/manipulation), the quality of engagement data, trust in inference providers; the litepaper mentions the vulnerability of single‑prompt scoring and the motivation for the multi‑LLM approach.

Project status and timeline

  • Current status: active (public website and playground).
  • Key releases/milestones: —.
  • Roadmap: described in the litepaper (Roadmap section).

Ecosystem/integrations

  • Integrations: web playground; Chromia is mentioned; inference providers (Hyperbolic/Groq) and X/Twitter engagement metrics (per the litepaper).
  • Partners: —.
  • Where it’s actually used: the playground and the “evaluate reply tweet” scenario (Tier‑1).

AI component of the project

  • What is real AI here? a multi‑LLM “LLM‑as‑a‑judge” system that evaluates AI agents’ responses on several criteria and aggregates a score (Tier‑1 litepaper).
  • What is truly onchain? storage of evaluation results in Chromia (per the litepaper) + the $EVAL token (ERC‑20) on the Base network.
  • Why is a token needed? litepaper: payment for evaluation reports under a subscription model in $EVAL.
  • Red flags (based on the information found): dependence on inference providers and on the quality of engagement signals; absence of public details about the economics of incentives/slashing; the risk of “judge hacking” with insufficient diversity/protection of the evaluators.

DecAI Fit component summary

  • DecAI Fit: 5/6 — high alignment with the Decentralized AI thesis
Component Level Score
AI reality high 3
Onchain necessity high 3
Token necessity low 1
Evidence Tier‑1 present 3
Details

AI reality

AI_0 — AI intent (Tier‑1 roadmap) (1/0)

  • AI_0 (decision): 0
  • Claim (one line): AI is described as a working part of the product (evaluation of AI agents), not only as intent/roadmap.
  • Evidence: Tier‑1 — https://www.evalengine.ai/ ; https://www.evalengine.ai/litepaper
  • Why sufficient / why not (one line): AI_0 adds no signal when Tier‑1 already describes a functioning AI component.
  • Conflict note (if any): —

AI_1 — what the AI does (1/0)

  • AI_1 (decision): 1
  • Claim (one line): The system evaluates AI agents’ responses via multi‑LLM judges and produces a composite score by quality criteria.
  • Evidence: Tier‑1 — https://www.evalengine.ai/litepaper ; https://www.evalengine.ai/
  • Why sufficient / why not (one line): Tier‑1 describes the purpose and mechanics of the evaluation (LLM‑as‑a‑judge + weighted scoring).
  • Conflict note (if any): —

AI_2 — demos/examples (1/0)

  • AI_2 (decision): 1
  • Claim (one line): There is a public playground (Evaluate Reply Tweet) for evaluating response quality.
  • Evidence: Tier‑1 — https://www.evalengine.ai/playground/tweets ; https://www.evalengine.ai/
  • Why sufficient / why not (one line): This is a reproducible demo surface that allows running an evaluation via the web interface.
  • Conflict note (if any): —

AI_3 — metrics/quality (1/0)

  • AI_3 (decision): 1
  • Metrics: composite score (weighted average by criteria); latency and model metrics/criteria (accuracy/creativity/truthfulness/engagement) are claimed.
  • Context (version/date/how measured): description of the mathematical model and examples of provider latency in the litepaper.
  • Evidence: Tier‑1 — https://www.evalengine.ai/litepaper
  • Why sufficient / why not (one line): Tier‑1 contains quantitative elements of the evaluation (formulas, criteria, a latency example).
  • Conflict note (if any): —

AI_4 — tech description/architecture/versions (1/0)

  • AI_4 (decision): 1
  • Claim (one line): The litepaper describes the system architecture (ingestion layer, evaluation engine, storage & smart contract layer, engagement module) and the process flow.
  • Evidence: Tier‑1 — https://www.evalengine.ai/litepaper ; https://eval-engine.gitbook.io/eval-engine
  • Why sufficient / why not (one line): There is a Tier‑1 tech description of the architectural components and the data flow.
  • Conflict note (if any): —

Onchain necessity

OC_1 — contracts/addresses/chain (1/0)

  • OC_1 (decision): 1
  • Chain: Base (token) + Chromia (storage, per the litepaper)
  • Addresses/identifiers: token contract 0xdd78523217390bb0d49c7601e7e54c36d71622f0
  • Evidence: Tier‑1 — https://basescan.org/token/0xdd78523217390bb0d49c7601e7e54c36d71622f0 ; https://www.evalengine.ai/litepaper
  • Why sufficient / why not (one line): There is an onchain artifact (token contract) and a Tier‑1 description of onchain storage of results (Chromia).
  • Conflict note (if any): —

OC_2 — onchain beyond just a token (1/0)

  • OC_2 (decision): 1
  • What useful logic is onchain: storage of evaluation results as a public immutable log in Chromia (gas‑free relational blockchain), plus “double‑signed” transaction confirmation (client + engine) in the description.
  • Evidence: Tier‑1 — https://www.evalengine.ai/litepaper
  • Why sufficient / why not (one line): Onchain is used as a layer of provability/auditability of results, not only as token issuance.
  • Conflict note (if any): —

OC_3 — onchain is critical (1/0)

  • OC_3 (decision): 1
  • “Replace the blockchain” test: without onchain, the key property “verifiable results / immutable history” is lost: it is impossible to verify trustlessly that the score and the evaluation history have not been tampered with.
  • Evidence: Tier‑1 — https://www.evalengine.ai/litepaper
  • Why sufficient / why not (one line): The litepaper formulates on‑chain storage as the basis of transparency/verifiability.
  • Conflict note (if any): —

Token necessity

TK_0 — no token (1/0)

  • TK_0 (decision): 0
  • Claim (one line): The project has a token, $EVAL (ERC‑20), on the Base network.
  • Evidence: Tier‑1 — https://basescan.org/token/0xdd78523217390bb0d49c7601e7e54c36d71622f0
  • Why sufficient / why not (one line): The Explorer confirms the token contract.
  • Conflict note (if any): —

TK_1..TK_4 (1/0)

TK_1 — payment for the service (1/0)

  • TK_1 (decision): 1
  • Claim (one line): Litepaper: each evaluation report is paid for with the $EVAL token via a subscription model.
  • Evidence: Tier‑1 — https://www.evalengine.ai/litepaper
  • Why sufficient / why not (one line): This is a direct Tier‑1 description of the token‑as‑payment mechanics.
  • Conflict note (if any): —

TK_2 — incentives/collateral/penalties (1/0)

  • TK_2 (decision): 0
  • Claim (one line): Protocol incentives/collateral/penalties requiring the $EVAL token are not established per Tier‑1.
  • Evidence: Tier‑1 — https://www.evalengine.ai/litepaper
  • Why sufficient / why not (one line): The litepaper describes payment and conversion to $CHR, but does not describe slashing/staking mechanics as a mandatory part of the protocol.
  • Conflict note (if any): —

TK_3 — governance (1/0)

  • TK_3 (decision): 0
  • Claim (one line): Governance use of the token $EVAL is not established per Tier‑1.
  • Evidence: Tier‑1 — https://www.evalengine.ai/litepaper
  • Why sufficient / why not (one line): The current litepaper has no specification of a DAO/voting.
  • Conflict note (if any): —

TK_4 — token is replaceable (1/0)

  • TK_4 (decision): 0
  • Claim (one line): The replaceability of the token is not established per Tier‑1.
  • Evidence: Tier‑1 — https://www.evalengine.ai/litepaper
  • Why sufficient / why not (one line): There is a payment model in $EVAL, but there is no formal test “it can be replaced with X without loss of meaning”.
  • Conflict note (if any): —

Evidence (table)

Tier Source What it proves
Tier‑1 https://www.evalengine.ai/litepaper Architecture, onchain storage of results (Chromia), score mathematics and token utility (subscription model in $EVAL).
Tier‑1 https://www.evalengine.ai/ Description of the product and links to the playground/benchmarking/analytics.
Tier‑1 https://www.evalengine.ai/playground/tweets Demo surface: evaluation of response quality (Evaluate Reply Tweet).
Tier‑1 https://basescan.org/token/0xdd78523217390bb0d49c7601e7e54c36d71622f0 Onchain fact: the $EVAL ERC‑20 token on the Base network, max total supply 1B.
Tier‑1 https://eval-engine.gitbook.io/eval-engine Tier‑1 docs: overview of the “decentralized framework for evaluating AI agents” and Chromia storage (context).
Tier‑2 https://coinmarketcap.com/currencies/eval-engine/ Reference token profile (not proof of utility).

No comments to display

Back to top