Simple. It doesn't compare to TOR or I2P.
Not only because the protocol stack is too heavy and the spec isn't feasible at all (which is why they haven't even started programming it yet. Been like a week), but because it's all cons as stated in these previous issues:

Technically:

• Betanet, from what I understood, is trying to be faster and better security wise than Tor and I2P combined. That's just not possible and I'm done with the tor/i2p slander in the comments under his video.
  We can notice that as his protocol stack (to strengthen security) include:

path-aware routing (SCION) at the path layer, HTX cover-transport tunneled over TCP-443 or QUIC-443 with an outer origin-mirroring TLS handshake and an inner Noise XK handshake (X25519 by default, hybrid X25519+Kyber768 mandated later), ChaCha20-Poly1305 AEAD + HKDF-SHA256 for symmetric protection, Ed25519 signatures, a libp2p-like overlay and rotating rendezvous DHT for peer discovery (mDNS / BLE / Onion v3 lists / IPFS fallbacks), optional Nym-style mixnet hops for added anonymity, Cashu federated e-cash + Lightning for payments, a 2-of-3 cross-chain alias ledger for naming/finality, plus cover connections, PoW-bound bootstrap, adaptive HTTP/2/3 emulation and many operational rules (rekeying, padding, calibration).

(^^^don't waste ur time reading the protocol list, you get the point lol).
Tor and I2P stack is more narrower:

Tor uses TCP-based three hop circuits with layered onion encryption (each relay removes one layer to preserve anonymity). Relay discovery and path selection rely on directory authorities and a signed consensus. It employs cell based framing, congestion control, and traffic shaping to resist analysis. Tor prioritizes proven, VERY WELL understood anonymity tradeoffs.

I2P relies on garlic routing and separate inbound/outbound tunnels, built per destination. It runs over NTCP2 (TCP) and SSU2 (UDP), bundling messages to resist traffic correlation. Peer discovery and routing rely on NetDB and floodfill routers. Unlike global path-aware systems or DHTs with PoW (proof of work), I2P emphasizes persistent tunnels and local router participation. It doesn't natively support QUIC or integrated payment protocols.
You can notice that despite the narrow protocol stack of tor and i2p, they're still quite slow in performance (if u call tor/i2p slow that's genuinely a skill issue). Betanet would only be fast on paper.

(The Raven Team thinks slapping QUIC and path selection will automatically make the protocol on top of their stack will automatically make it faster, next sprint they’ll add Redis for speed lmao.)

Simply:

• The Raven Team's codebase is all vibe coded. You can see that from this readme, they barely perfect or correct any imprecisions (There are a ton of evidence supporting this, I'm not going to list them all. As stated in, Betanet version confusion #44 , "Version 1.X" is a common term for AI models, there's no specific version of it. It's hard to even define its architecture, at this point it doesn't even look like a spec but a scifi thriller script. Ok, lol, jokes aside, just because it's a spec it doesn't excuse for missing documentation (like separate RFC, IETF (e.g. https://datatracker.ietf.org/doc/html/draft-grothoff-iesg-special-use-p2p-names-04), and formal models use like TLA+ / Coq)).

pros:
The only pro it has is the $40K funding it has received through a crypto token ($BETANET), cough cough though with no progress at all.

(if u slander tor or i2p i slander u)