Abliterated bf16 safetensors base of 0xSero/Qwen3.5-88B - 0xSero's ~30% MoE expert-prune (REAP) of Qwen/Qwen3.5-122B-A10B, taking the model from 122B down to ~88B parameters while keeping the A10B active-expert budget and the qwen35moe architecture. Robinson Labs then abliterated the pruned model in-house. This repo is the full-precision master, in safetensors.

This is the bf16 base that the RobinsonLabs/Qwen3.5-122B-A10B-REAP-30-abliterated-GGUF quant ladder was quantized from. If you want a ready-to-run quant, use that repo. This repo is the master for further surgery (re-abliteration, LoRA merge, fine-tune) and for rolling your own quants.

Unlike the full 122B sibling, this REAP variant has no Multi-Token Prediction (MTP / NextN). The real model is a clean 48-layer qwen35moe - standard single-token prediction.

Disclosure

This model is abliterated - the hard-refusal reflex on adult / creative content has been reduced via single-direction weight orthogonalization. Harm guardrails are retained by design: self-harm prompts still redirect to help (e.g. 988), and it is not intended to assist genuine wrongdoing. This is a v1, partial abliteration; capability is preserved. Tagged not-for-all-audiences. Use responsibly - you are responsible for your use. License inherited from the base model: Apache-2.0.

Method

Expert prune (REAP): ~30% of the MoE experts removed by 0xSero's REAP method, 122B -> ~88B params, qwen35moe arch, A10B active budget retained. Done upstream, in 0xSero/Qwen3.5-88B.
Abliteration: single mid-layer refusal direction removed via weight orthogonalization on the bf16 pruned base; routers preserved. No MTP/NextN block exists in this variant.
Format: safetensors, sharded, with config + tokenizer + index. No vision tensors, no MTP head.

Files

Format	Precision	~Size	Notes
safetensors (44 shards)	bf16	~171 GB	abliterated REAP base; `qwen35moe`, 48 layers, no `nextn`

The model is 48 transformer layers (block_count=48), qwen35moe architecture, ~88B params with an A10B active-expert budget. The upstream config declares a phantom nextn (MTP) layer that carries no weights - downstream GGUF converters should treat this as a clean 48-layer model (block_count=48) and ignore the phantom head. (This is the trap that produced the single-token GGUF sibling.)

Quants

GGUF quants (Q6_K down to IQ2_M, imatrix-weighted, 48-layer / no MTP) are published at RobinsonLabs/Qwen3.5-122B-A10B-REAP-30-abliterated-GGUF.

Provenance

Qwen3.5-122B-A10B (Apache-2.0) -> REAP-30 expert-prune (0xSero, repo "Qwen3.5-88B") -> abliterated bf16 (Robinson Labs). This safetensors repo is the abliterated bf16 master; the GGUF ladder is quantized from it.