in the Sherlock Symbiotic contest (Sep 2025) — competing against dozens of expert human auditors.
Source ↗AI SMART CONTRACT AUDITS — AND THE CODE AROUND THEM
Find Smart Contract Vulnerabilities Before Attackers Do
Deeper than a scanner. Faster than a manual audit. AI security for smart contracts — Solidity, Vyper, Rust, or any other language — and for everything your protocol depends on: the circuits, the node, the backend that touches your keys.
Trusted by 1inch, Lido, and Pessimistic SecurityTop-6 finish in a Sherlock audit contest
TRUSTED BY SECURITY TEAMS AT
YOUR CONTRACTS — AND EVERYTHING AROUND THEM
Smart contractsZK circuits & cryptographyNode clients & infrastructureOff-chain services & keepersAny other code
Solidity · Vyper · Rust · Circom · Go · C++ · TypeScript · Python — and any other language
One language-agnostic engine behind them all — proof published domain by domain.
PROOF
The numbers behind the name.
Every number below comes from public, verifiable smart contract audits and benchmarks — links included.
on the Crestal Sherlock contest in an independent blind pilot: 7/7 judge-adjudicated issues, each with a PoC.
Source ↗precision — the best among AI auditors across three judge-adjudicated contests in the same pilot.
Source ↗accuracy on CTFBench, our open smart contract benchmark (methodology published on ethresear.ch).
Source ↗vulnerability classes, curated from a reference book of 20,000 real-world smart contract vulnerabilities.
Source ↗free credits on signup — no card required. Enough to audit a typical ERC-20 end-to-end.
Source ↗WHY SAVANT CHAT
Deeper. Faster. Trusted.
Deeper than a scanner
Static analyzers pattern-match known bug shapes. Our multi-agent stack runs thousands of specialized LLM calls across 200+ vulnerability classes — each distilled from real-world exploits in a 20,000-case smart contract reference book. On Advanced and Pro tiers, a critic subagent builds a proof-of-concept for every finding before it reaches your dashboard.
Faster than a manual audit
A manual smart contract audit costs $40K–$100K and takes 3–38 days. Savant Chat audits a typical codebase in 10–30 minutes, pay-as-you-go from $0.07/line — on every commit via GitHub and GitLab CI.
Trusted before the human audit
1inch, Lido, Pessimistic Security, OXORIO, and other security teams run Savant Chat first, so their auditors' time goes to the bugs AI can't yet see. Second pair of eyes — not the last one.
COVERAGE
Every major contract language. And everything around your contracts.
Smart contracts are where we publish benchmarks and contest results. The engine underneath is language-agnostic — the same agents audit the node your chain runs on, the circuits your privacy depends on, and the backend that touches your keys.
Smart contracts
Sherlock Top-6 · CTFBench 87–95% · flagship domain
SolidityEthereum + EVM L2s
VyperCurve-style DeFi
RustSolana · NEAR
Any contract languageMove · Cairo · FunC · …
Cross-function reentrancyRead-only reentrancyOracle manipulationGovernance attacksMEV extractionFlash-loan vectorsCross-chain messagingEconomic exploitsSandwich attacksBlock-timestamp manipulationL2 bridge vulnerabilities200+ classes total
ZK circuits & cryptography
Circom · Halo2 · Noir · arkworks · …
Under-constrained circuitsSoundness gapsNonce reuse & key handlingZK-privacy leaksSignature malleability
Nodes & infrastructure
Rust · Go · C++ · …
Consensus edge casesP2P & DoS vectorsState-sync corruption
Off-chain code
TypeScript · Python · Go · …
Authorization bypassInjection & deserializationSupply-chain risksBusiness-logic edge cases
The same engine behind every domain above — and any other code you upload. See the full matrix and upload formats →
COMPARISON
Savant Chat vs the alternatives — at a glance.
Head-to-head on smart contract audits.
| Criterion | Slither/Mythril | Manual audit firm | AuditAgent | Octane | Savant Chat |
|---|---|---|---|---|---|
| Coverage | Syntactic patterns only (no logic/economic reasoning) | Depends on auditor | Broad LLM + static + dynamic | LLM agents + fuzz detectors | 200+ classes, multi-agent, RAG reference book |
| Time to result | Minutes | 3–38 days | Hours–days | Minutes–hours | Minutes (10–30 for typical code) |
| Cost (typical DeFi protocol) | Free | $40K–$100K (Sherlock) | From $199/mo subscription | Invitation-only, sales-led | $0.07–$0.50/line, $75 free credits |
| Public benchmark | ~46–54% recall (Augment Code summary) | n/a | 67% recall on EVMBench (post-validation) | Monad contest #1 of 1,600 | 87–95% on CTFBench (our open benchmark); 100% recall on Crestal (judge-adjudicated) |
| False positives | Very high (developers ignore warnings) | Low (human-reviewed) | 75% FP reduction via validation phase | Lower — validates every finding (vendor claim) | Critic subagent PoCs every finding (Advanced & Pro); "didn’t produce a single clear false positive" — Pessimistic Security |
| Language support | Solidity | Varies | Solidity, some Solana, Starknet | Language-agnostic | Language-agnostic — any code: smart contracts (Solidity · Vyper · Rust), ZK circuits, nodes, backends |
| CI/CD | Mostly no | No | Yes | Yes (GitHub/GitLab) | Yes (GitHub & GitLab CI) |
Roughly 1–3% of the cost of a manual DeFi audit — with a public contest record.
WHAT SECURITY TEAMS SAY
Used before senior auditors spend their time.
1inch
1inch is evaluating @savantchat, aiming to catch and fix issues early — in development, not in production — in minutes, not days — at a fraction of the cost of a human audit — finding errors humans overlook.
Vasiliy Shapovalov
This checks with my impression, @savantchat does a great job of filtering false positives while finding issues. It's not an audit replacement — as faster, cheaper tool its place in developer pipeline is closer to an internal review or a heavier linter run.
Pessimistic Security
We recently tested savant.chat and were pleasantly surprised! It correctly identified several findings on our test contract and didn't produce a single clear false positive. This is the first genuinely useful security tool we've come across in quite a while.
OXORIO
We recently tested @savantchat — an AI-powered auditing tool — on a real DeFi project (~3k SLOC), previously audited by multiple top firms. The question: can AI surface anything meaningful post-audit? Spoiler: it can. And it made us rethink how automation can augment human…
Petr Korolev
I just put savant.chat to the test on a complex contract — and wow, what a game-changer! It uncovered a critical issue that many seasoned auditors overlooked, proving its ability to boost audit quality. This tool is set to redefine smart contract security!
PRICING
Pay as you go. Priced per line.
$75 in free credits on signup — no card required. Credits never expire.
Lite
$0.07/line
Efficient models. CI-friendly quick pass.
- Severity-ranked findings
- Confidence score on each issue
- GitHub/GitLab CI
Advanced
$0.12/line
Comprehensive analysis. The default for pre-audit runs.
- Severity-ranked findings + confidence score
- PoC-validated by critic subagent
- GitHub/GitLab CI
- Report export
Pro
$0.50/line
Highest-quality models for critical releases.
- Highest-quality model pass
- PoC-validated findings
- Report export
- Critical release review
Account tiers: Basic $250/mo · Pro $2,500/mo · Enterprise custom.
The per-line pricing above applies to any code — no quote needed. Auditing a very large scope, like a full node client or a monorepo? Contact us for a flat scoped quote.
FAQ
Honest answers to hard questions.
Can AI really audit a smart contract?
Yes — for the 200+ vulnerability classes our multi-agent stack covers, with PoC validation of every finding on Advanced and Pro tiers. No — for novel economic designs where the auditor needs to reason about incentives and cross-protocol invariants. The right model is AI first, human second.
Does Savant only audit smart contracts?
No. Smart contracts are our first and best-proven domain — that's where our contest results and benchmarks live. But the engine underneath is language-agnostic: the same multi-agent pipeline audits ZK circuits, node clients, off-chain backends — any code you upload. If you're here for an EVM audit, that's still what we do most — and prove best. See Supported Languages in our docs for the full matrix.
How is this different from Slither or Aderyn?
Static scanners pattern-match known syntactic shapes. Savant Chat traces value flows, models attacker incentives, covers 200+ vulnerability classes, and on Advanced and Pro tiers uses a critic subagent to PoC each finding before it reaches you.
How is this different from Semgrep, CodeQL, or Snyk?
Those tools match known patterns. Savant runs hundreds of specialized LLM agents in parallel — one per vulnerability class — tracing data flows, modeling attacker behavior, and validating findings with a proof-of-concept critic. Use both: pattern scanners for hygiene, Savant for bugs that need reasoning.
What about false positives?
Every finding has a severity rating and confidence score. On Advanced and Pro tiers, a separate critic subagent builds a proof-of-concept exploit before the finding surfaces; if the PoC fails, the finding is discarded.
Do I still need a human audit?
Yes. Novel economic-design invariants are still where AI misses most — that's why Savant Chat is the second pair of eyes, not the last one. Run it before the manual audit so the human auditor's time goes to the bugs AI cannot yet see.
Is my code private?
Your code is processed securely and isn't stored or shared beyond what's needed to run your audit. We work with trusted AI providers under strict data-handling agreements.
What does it cost?
Pay as you go per line of code: Lite $0.07/line, Advanced $0.12/line, Pro $0.50/line. $75 in free credits on signup, no credit card.
$75 free. No card required.
Run a real audit on your codebase today — results in minutes.
Start your first audit