---

library_name: gguf

license: mit

pipeline_tag: text-generation

base_model:

• XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B
• ornith-ai/Ornith-1.5-9B

language:

• en
• zh

tags:

• qwen3.5
• slerp
• merge
• coding
• agentic
• multimodal
• vision

---

Qwen3.5-9B-Ornimo-SLERP (GGUF)

GGUF quantizations of a Qwen3.5-9B hybrid-architecture SLERP merge: the

self-improvement RL tower of

Ornith-1.5-9B

blended into the agentic-SFT shell of

MiMo-V2.6-Distill-Qwen-9B.

An ornith (a bird) and a MiMo walk into a merge — Ornimo.

Files

| file | size | notes |

|---|---|---|

| Qwen3.5-9B-Ornimo-SLERP-Q4_K_M.gguf | 5,629,105,024 B (5.63 GB) | 4-bit K-quant; recommended for everyday use |

| Qwen3.5-9B-Ornimo-SLERP-F16.gguf | 17,920,693,120 B (17.92 GB) | unquantized F16 text weights (converted from the bf16 checkpoint) |

| Qwen3.5-9B-Ornimo-SLERP-mmproj-f16.gguf | 921,705,024 B (922 MB) | vision tower + projector (MiMo's, byte-exact), F16 |

That is the complete list. This repo contains GGUF weights only — no

safetensors (those live in the sibling checkpoint repo) and no imatrix

(importance matrix): the Q4_K_M was produced by a plain, calibration-free

llama-quantize … Q4_K_M run with no calibration dataset. If you prefer an

imatrix-calibrated quant, generate an imatrix from the F16 GGUF and

requantize — the F16 is provided exactly for that.

What is this, really?

Nobody trained anything here. Both parents are fine-tunes of the same

Qwen3.5-9B backbone (32 hybrid layers — 24 linear-attention + 8

full-attention, hidden 4096, vocab 248320, 262K context) with the

identical tokenizer (248,044 tokens, same ordering), so the text towers

align 1:1. The merge is a whole-tensor SLERP (t=0.5) of all 427 text

tensors, with MiMo's vision tower, projector, tokenizer, chat template and

configs kept byte-exact. Ornith's optional multi-token-prediction (mtp.*)

block is not part of the merge — it is trained against Ornith's own

hidden space and would be miscalibrated on a blended one.

Usage

Both GGUFs use the Qwen3.5 hybrid arch (qwen35 in llama.cpp naming) and

run on llama.cpp b393+.

Text only:

llama-cli -m Qwen3.5-9B-Ornimo-SLERP-Q4_K_M.gguf \
  -c 131072 -t 16 -cnv --repeat-penalty 1.1

Vision (image understanding) with the projector:

llama-mtmd-cli -m Qwen3.5-9B-Ornimo-SLERP-Q4_K_M.gguf \
  --mmproj Qwen3.5-9B-Ornimo-SLERP-mmproj-f16.gguf \
  --image <image.png> -p "What is in this image?" -c 131072 -t 16

• -c context: the model supports up to 262144 natively; only the 8

full-attention layers cache KV, the 24 linear-attention layers keep

fixed-size state.

• Thinking mode follows MiMo's chat template

(chat_template_kwargs: {"enable_thinking": true} in OpenAI-compatible

servers; /no_think in the prompt for a direct answer).

• --repeat-penalty 1.1 is recommended for very long reasoning chains.

Why this merge was safe (deviation analysis)

Before merging, both parents were compared tensor-by-tensor

(760 shared tensors):

| check | result |

|---|---|

| tokenizer vocab ordering | identical (248,044 tokens, same order) |

| median tensor cosine (text) | 0.993–1.000 per category |

| min tensor cosine | 0.986 (layers.22.linear_attn.in_proj_b) |

| tensors below 0.98 cosine | 0 |

| gating tally (lo=0.90, hi=0.995) | 293 full-SLERP / 134 gated / 0 anchored |

| lm_head row cosine | min 0.930, no row below 0.8 — no clashing token families |

| vision towers | cos 1.000 median — near-identical, copied byte-exact |

| NaNs | 0 |

Because tokenizers are identical and every output row agrees, the

unembedding (lm_head) is merged along with everything else — no

anchor-side lm_head exception.

Verified behavior (CPU smoke checks, not a benchmark suite)

• structural verify: all 4 output shards recomputed against an independent

SLERP implementation (1-ulp tolerance) — passed

• Q4_K_M via llama.cpp: 17 × 24 = 408 plus a clean one-line Python

function, no repetition loops, ~13 tok/s generation on 24 CPU threads

(~100 t/s prompt processing)

Everything beyond these checks is uncharted; expect surprises and

hallucinations in unknown proportions.

Credits

• XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B

— the anchor: shell, vision tower, projector, tokenizer, chat template

(MIT).

• ornith-ai/Ornith-1.5-9B

— the RL-self-improvement parent whose text tower is blended in (MIT).

• Qwen3.5-9B — the shared backbone both parents fine-tune

(Apache-2.0, Copyright 2026 Alibaba Cloud).