Overview
Succinct.xyz builds tools and protocols to improve blockchain efficiency and cross-chain compatibility. It verifiably proves Bitcoin block headers to let light clients confirm data without downloading the full chain, and provides Telepathy, a zkSNARK-based interoperability protocol for Ethereum that enables secure cross-chain communication with succinct proofs. It also offers a gas-efficient on-chain light client for Ethereum’s Proof of Stake powered by zkSNARKs, reducing the resources needed to verify transactions. The company aims to help developers, enterprises, and blockchain enthusiasts achieve better interoperability and efficiency across networks using practical cryptographic tooling.
Work here?Claim your company
About Succinct
Industries
Company Size
11-50
Company Stage
Series A
Total Funding
$55M
Headquarters
San Francisco, CA
Founded
2022
Get referred to Succinct
See people who can refer or advise you
Simplify's take
What believers are saying
- Succinct launched ZCAM on April 23, 2026, extending cryptography beyond blockchain into media provenance.
- The October 2026 SP1 patch shows active security response after the verifier bug.
- The 2026 ZK Industry Report positions Succinct with governments, banks, and Big Tech.
What critics are saying
- On October 11, 2026, GHSA-p7v2-jjv3-8p8f let native Rust verifiers accept invalid proofs.
- August 5, 2026 unlocked 208.33 million PROVE, crushing liquidity and signaling weak token economics.
- A fatal SP1 soundness flaw would undermine every chain integration and kill trust.
What makes Succinct unique
- SP1 Hypercube proved 99.7% of Ethereum mainnet blocks in under 12 seconds on 16 GPUs.
- Base's May 4, 2026 Azul upgrade uses SP1 for $7.4 billion deposits.
- Chainlink's September 28, 2026 CCIP 2.0 testnet adds ZK verification.
Help us improve and share your feedback! Did you find this helpful?
Funding
Total Funding
$55M
Above
Industry Average
Funded Over
2 Rounds
Industry standards
Benefits
- Health Insurance
- Dental Insurance
- Vision Insurance
- Company Equity
- Meal Benefits
- Conference Attendance Budget
Growth & Insights and Company News
6 month growth
-6%1 year growth
0%2 year growth
3%Anthropic releases FlashProve, zkSNARK prover for LLM inference. Anthropic's FlashProve generates zkSNARK proofs of LLM inference at 110 tokens/sec, enabling cryptographic audit of model output without weight access. Anthropic, in collaboration with researchers from EPFL, MIT, StarkWare, RESI, and Succinct, has released FlashProve, an open-source system for provable inference built on zero-knowledge succinct arguments of knowledge. The system allows a model provider to commit to a specific set of weights and then generate a short certificate for each response proving that the committed model produced it. For a 400-billion-parameter mixture-of-experts model on a single H100 GPU, the team reports proof generation at around 110 tokens per second, with proof sizes of roughly 19 MB per 1,024-token request. Why it matters. The problem FlashProve addresses is practical and specific: today, a user querying an AI model has no way to verify, from the output alone, that the response was generated by the model the provider claims is serving it. Anthropic identifies three concrete use cases. Safety evaluators testing models before and after release can confirm the model under test is the one being deployed. Clients using third-party routing services can verify the tokens came from the model they paid for. In a threat model where an attacker or a more capable model substitutes itself for an authorized one, valid proofs bound to a specific weight commitment would be impossible for the impostor to produce. Anthropic connects the work to a goal in its Frontier Safety Roadmap to produce a prototype of provable inference by September 30, 2026, which it says was met through a separate but related project that signs inference traffic and recomputes it on a trusted server. How the zkSNARK protocol works. FlashProve is a software-only system built around a cryptographic primitive called a zkSNARK, which Anthropic describes as "uniquely fit for provable inference" due to its strong hiding and efficiency properties. The protocol operates in two stages. First, the provider commits to a specific model. For each subsequent response, the provider generates a short certificate demonstrating that the response was produced by that committed model. The certificate has two defining properties. It is zero-knowledge: it reveals nothing about the model weights beyond what the correctness of the computation already implies. It is succinct: verification is exponentially faster than re-running the computation, allowing a user to check a proof in seconds on a CPU. The system uses 4-bit quantized weights and integer arithmetic during the prefill stage. Anthropic says the design relies on well-established principles for cryptographic proof systems while applying state-of-the-art tools from the cryptography and security literature in novel ways; the team attributes these applications to Claude, the model used in the development process. The protocol specification document included in the release is explicitly described as Claude-written. Because an AI model participated in building the system, the team addresses the trust question in the cryptographic guarantees directly. Anthropic acknowledges that Claude "often makes decisions that are hard to evaluate, especially in large complex systems," and counters this by expressing the top-level security guarantees as auditable statements in Lean, a formal verification language. The release includes Lean proofs for the top-level mathematical claims alongside an interactive visualizer of the theorems to aid external auditing. Hiding the model architecture is explicitly not a goal of the system; the proofs hide only the weights. Benchmark performance. The benchmark model is Llama 4 Maverick, a 400-billion-parameter mixture-of-experts architecture with 4-bit quantized weights, running on a single H100 GPU. The team reports the following figures: * Proof generation speed: around 110 tokens per second, described as "over 100 tokens/second" in the introduction * Proof size: approximately 19 MB per 1,024-token request * GPU overhead: roughly 70 to 95 times the GPU time that vLLM, a common open-source inference library, requires to process prompts of the same length Anthropic characterizes prior open-source systems for provable inference as "orders of magnitude slower" but does not provide specific baseline numbers in the post. What was released and how to access it. The team released five artifacts under the Apache 2.0 license, covering code, Lean proofs, and documentation: * The blog post outlining the system and its potential applications * A technical report explaining implementation details * A Claude-written protocol specification document * Executable code containing a prover, a fast verifier in C++, and a provably correct verifier in Lean * Lean proofs for the top-level mathematical statements, with an interactive visualizer of the theorems to aid auditing The team states the goal is that others "look at what we have done, take it, and improve it," and describes the release as "intermediate progress, and not a completed research agenda." Intended audiences include developers, safety evaluators, regulators, auditors, and the research community. The authors listed on the release are Roy Rinberg, Alex Allain, Alessandro Chiesa, Miranda Christ, Rahul Dalal, Giacomo Fenzi, Ben Mayer, and Guy Weissenberg. Limitations noted by the team. Two constraints are stated explicitly. Hiding the model architecture is outside the system's scope. And for the pacing and regulation use case, Anthropic notes that while the proofs verify the hardware is serving inference, they "do not establish that the hardware is not, e.g., training on the side." References For more details, visit:
Introducing the 2026 ZK Industry Report. Today, Succinct and the Zero Knowledge Podcast are releasing the first ZK Industry Report, tracking the state of proving technology and its move into production across industries. What's inside. * A foreword from Dan Boneh, Professor of Computer Science at Stanford, on where proof systems are headed next. * The state of proving and research, from real-time Ethereum proving to the shift toward quantum-resistant systems. * ZK in production at governments, Big Tech, and global banks, from national digital ID to tokenized bank deposits. * What comes next, and why AI, regulation, and quantum are moving ZK towards critical infrastructure. Get the report. Whether you're building products, managing risk, or tracking where infrastructure is headed, the report is a single snapshot of where ZK stands in 2026. Thank you to the Zero Knowledge Podcast team for partnering with Succinct on this report.
Succinct patches SP1 bug allowing invalid proofs to pass verification. Phemex News 2026/10/23 07:21 Succinct has released a critical patch for its SP1 zero-knowledge proof system after discovering a bug that could allow invalid proofs to pass verification. Native Rust verifiers accepting untrusted proofs must upgrade to version 6.8.1 immediately, the company said. Onchain Groth16 and Plonk verification via Solidity remains unaffected by the vulnerability. Disclaimer: The content provided on Phemex News is for informational purposes only. Phemex do not guarantee the quality, accuracy, or completeness of the information sourced from third-party articles. The content on this page does not constitute financial or investment advice. Phemex strongly encourage you to conduct you own research and consult with a qualified financial advisor before making any investment decisions.
Succinct and Chainlink roll out ZK security for institutional cross-chain transfers. * Guido Battigelli * Published: September 28, 2026 * 9:46 pm * Updated: September 28, 2026 * 9:46 pm Table of Contents Succinct introduced a new zero-knowledge (ZK) verification module for Chainlink CCIP 2.0, the cross-chain interoperability protocol from Chainlink, with the goal of strengthening the security of high-value transfers between blockchains. The integration, currently available on testnet on the Ethereum Sepolia to Arbitrum Sepolia lane, allows any asset issuer, bank, or protocol to require a cryptographic proof of Ethereum's finality before funds move to another EVM network. The module is built on SP1 Helios, the ZK-powered Ethereum light client developed by Succinct, which generates verifiable proofs directly on the destination chain. In a configuration with this Cross-Chain Verifier enabled, both Chainlink's default verification path and the ZK path must be confirmed before the transaction is executed, adding independent layers of defense without replacing the existing architecture. The adjusted stablecoin volume surpassed $8.8 trillion in the first half of 2026, while tokenized real-world assets reached $38 billion, more than 400% above early 2025 levels. Bridge hacks have already accumulated losses exceeding $3.6 billion, making cryptographic verification a priority for institutional capital operating on-chain, and Succinct aims to deliver that level of security. Disclaimer: Crypto Economy Flash News are based on verified public and official sources. Their purpose is to provide fast, factual updates about relevant events in the crypto and blockchain ecosystem. This information does not constitute financial advice or investment recommendation. Readers are encouraged to verify all details through official project channels before making any related decisions. Technology TL;DR: ENS Labs and GLEIF are exploring linking ENS names with verifiable legal entity identifiers (vLEI) to authenticate institutions onchain. The proposal includes an ENS Bitcoin News TL;DR Researchers published a 56-page specification for Zcash-style private Bitcoin transfers, hiding sender, receiver, and amount. The protocol, called Shielded Bitcoin, uses zero-knowledge proofs and CryptoNews TL;DR Bitfinex Securities operates under AIFC regulation in Kazakhstan, a jurisdiction that extended its legal framework to tokenized instruments in October 2024. Hong Kong issued flash news Web3 cybersecurity and audit firm CertiK joins LF Decentralized Trust (LFDT), the Linux Foundation's open-source initiative focused on decentralized ecosystems. Ronghui Gu, CEO and co-founder flash news Tezos launched Quantumnet, its first experimental post-quantum testnet, developed by engineers from Nomadic Labs, Trilitech and other ecosystem contributors. Technology TL;DR European financial authorities warned that quantum computers could break the cryptography protecting blockchains before having any viable commercial use. According to Cryptoquant, approximately 6.9 Follow Crypto Economy on Social Networks Crypto Tutorials Crypto Reviews
Base adds ZK proofs to Base Azul with SP1. Base is partnering with Succinct to prove $7.4 billion in deposits with zero-knowledge proofs. Today, Succinct is announcing that Base is partnering with Succinct to bring zero-knowledge proofs to the Base chain. Base's Azul upgrade will use SP1 to prove $7.4 billion in deposits. This is a defining moment for Ethereum and its scaling roadmap. With Base upgrading to zero-knowledge, the largest layer 2 chains have validity proofs on their roadmap. Leading chains like Base use SP1 to generate their proofs, making SP1 critical infrastructure for chains that need speed and security. Proving Base with SP1. Base's current fault-proof system is optimistic and interactive: a batch of transactions is assumed valid unless someone challenges it within a 7-day window, slowing down withdrawals for $7.4B in assets. The upcoming Azul proof system upgrades Base to a multi-proof system combining TEE and ZK proofs as an intermediary step towards full ZK proving. When a ZK proof - generated by SP1 - and TEE proof back the same proposal, the Base chain unlocks three important properties: * Faster finality. Finality drops from seven days to one if a proposal to withdraw funds back to mainnet is backed by both a ZK and TEE proof. That means faster withdrawals and better capital efficiency. * Cryptographic security. Fault proofs rely on proposers to bond capital when submitting proposals. Azul's ZK path complements Base's existing security model with cryptographic guarantees that help prove correctness. * A path to Stage 2 decentralization. Reaching Stage 2 requires a way to detect proof system errors onchain. Azul accomplishes this through contradiction: ZK proofs are posted permissionlessly and override permissioned TEE proofs when the two disagree. The most advanced zkVM in production. Within the past year, several major chains have migrated from optimistic proof systems to ZK proofs. This is because zero-knowledge virtual machines (zkVMs) like SP1 have become increasingly performant and cheap. Generating ZK proofs for L2 blocks used to take hours and required writing bespoke circuits. SP1 lets teams write programs in standard Rust, compile to RISC-V, and generate ZK proofs of execution. With SP1, generating ZK proofs becomes fast, cheap, and production-ready for leading chains like Base. Over 35 customers, including rollups, bridges, exchanges, and DA layers use SP1 for its: * Performance. SP1 Hypercube is the first zkVM to deliver real-time proving for Ethereum, proving 99.7% of Ethereum mainnet blocks in under 12 seconds on 16 RTX 5090 GPUs. * Security. SP1 Hypercube is the first zkVM with complete formal verification of all 62 core RISC-V opcodes, completed with Nethermind Security and the Ethereum Foundation. * Trust. SP1 secures billions in TVL across the largest L2s and has generated millions of proofs on the Succinct Prover Network. $10B secured and counting. Base joins Optimism, Arbitrum, Polygon, Mantle, Celo, and more in implementing SP1 to secure over $10 billion in digital assets across major rollup ecosystems. The Ethereum rollup ecosystem is consolidating around ZK, with SP1 at the center.
There are no jobs for Succinct right now.
Find jobs on Simplify and start your career today
We update Succinct's jobs every few hours, so check again soon!Browse all jobs
About Succinct
Industries
Company Size
11-50
Company Stage
Series A
Total Funding
$55M
Headquarters
San Francisco, CA
Founded
2022
There are no jobs for Succinct right now.
Find jobs on Simplify and start your career today
We update Succinct's jobs every few hours, so check again soon!Browse all jobs