Home > ZKPs > Framework-Decision-Guide
Framework Decision Guide¶
Last modified: 2026-10-28
Evaluation criteria and a shortlist of ZKP frameworks for e-ID credential proofs (from the eid-privacy project).
Evaluation criteria
The columns below were evaluated in the eid-privacy project (GHE26-MS2). BL26-SoK-AnonCreds adds questions worth asking of any candidate:
- Is it actually zero-knowledge? Many systems built for blockchains target succinctness only, and some zkVMs are not ZK.
- Prover time on a phone, not proof size or verifier time, is the bottleneck for credentials (ZKP-on-Mobile).
- Security level: some fast results use reduced security (Longfellow: 86 bits).
- Setup and assumptions: trusted setup vs. transparent, pairing-based vs. hash-based (ZK-Families-Overview).
- Assurance: audits, known bugs, formal verification (Soundness-Bugs-and-Audits).
- Extensibility: tailored circuits are fast but hard to extend to new statements.
Framework shortlist
| Framework | Ease | Fast | PQ | Mobile | Maintained | Audited | Sign+Hash |
|---|---|---|---|---|---|---|---|
| Noir | Easy | Yes | No | Possible | Yes | No | Yes |
| Circom 2 | Medium | Unclear | No | WASM | Mixed | No | Hash only |
| gnark | Medium | Yes | No | No | Yes | Yes (9×) | Yes |
| RISC Zero | Medium | Mixed | Yes | Verifier | Yes | Yes | Yes |
| Docknetwork | Medium | Yes | Partial | Yes | Mixed | No | Yes |
| Ligero Prover | Easy | Partial | Yes | WASM | Yes | No | Yes |
| Halo2 | Hard | Unclear | No | Unknown | Some | No | Hash only |
| Spartan | Low-level | Yes | No | Likely | Mixed | No | Yes |
Background reading (recommended order)
- ZK Families Overview — understand the landscape
- Proof Systems — read the system your shortlisted framework is built on
- Intermediate Representations — understand what your circuit compiles to
- Resources — Thaler book for depth, Evolution of ZKPs for history