architectures · inclusionai

Bailing Hybrid (Ling 3.0)

Sparse mixture-of-experts backbone behind the Ling 3.0 family, with 128 experts and 8 routed per token, and a low-rank compressed key/value projection (kv_lora_rank 512) rather than a full per-head KV cache. Declared as model_type `bailing_hybrid`, architecture class BailingMoeV3.

Verify

What this registry has checked

Quarantine. We hold the publisher’s own published hash for these weights and have recorded it against this entry. We have not downloaded and hashed the weights ourselves, so the hash is the publisher’s claim, not our measurement.
Re-check this yourself → Source → Full record →

What is recorded

Publisherinclusionai
OriginChina
LicenceApache-2.0

Elsewhere in the registry

All architectures → Browse with filters → What will run on my hardware? →

Registry record as of 2026-08-14. Identity, licence and parameter count are properties of a release and do not move; tier can, and the verify link above re-checks it against the live registry rather than this page.