--- title: Ornith-1.0-9B-1M-GGUF canonical_url: "https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF" md_url: "https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF.md" repository: satgeze/Ornith-1.0-9B-1M-GGUF last_updated: 2026-07-06 license: mit pipeline_tag: text-generation tasks: - text-generation base_model: - deepreinforce-ai/Ornith-1.0-9B base_model_relation: quantized library_name: - gguf downloads: 1293 stars: 0 tags: - gguf - long-context - yarn - ornith - mtp - speculative-decoding - vision - llama.cpp - ollama --- # Ornith-1.0-9B-1M-GGUF > Ornith-1.0-9B-1M-GGUF - An open-source model by satgeze on ModelScope. Ornith-1.0-9B: 1M Context + MTP + Vision satgeze/Ornith-1.0-9B-1M-GGUF is a text-generation model on ModelScope. licensed under mit. derived from deepreinforce-ai/Ornith-1.0-9B. - **Repository**: satgeze/Ornith-1.0-9B-1M-GGUF - **License**: mit - **Tasks**: text-generation - **Base model**: deepreinforce-ai/Ornith-1.0-9B - **Tags**: gguf, long-context, yarn, ornith, mtp, speculative-decoding, vision, llama.cpp, ollama - **Downloads**: 1293 - **Stars**: 0 - **Last updated**: 2026-07-06 Source: https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF ---
Ornith # Ornith-1.0-9B: 1M Context + MTP + Vision [![GitHub](https://img.shields.io/badge/GitHub-aviary--1m%20(harness%20%2B%20method)-181717?logo=github)](https://github.com/satindergrewal/aviary-1m) [![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](https://opensource.org/licenses/MIT) [![Base](https://img.shields.io/badge/base-deepreinforce--ai%2FOrnith--1.0--9B-8A2BE2)](https://huggingface.co/deepreinforce-ai/Ornith-1.0-9B) [![Context: 1M](https://img.shields.io/badge/context-1%2C048%2C576%20tokens-8A2BE2)]() [![MTP](https://img.shields.io/badge/MTP-baked%20in-brightgreen)]()
**Mirrors:** [Hugging Face](https://huggingface.co/satgeze/Ornith-1.0-9B-1M-GGUF) | [ModelScope](https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF) (full quant ladder always available on ModelScope) [DeepReinforce's Ornith-1.0-9B](https://huggingface.co/deepreinforce-ai/Ornith-1.0-9B) (9B dense, Qwen3.5 family) with YaRN RoPE scaling **baked into the GGUF metadata** for a **1,048,576-token context window**, 4x the native 262,144, shipping with the MTP speculative-decoding layer baked in and a vision tower alongside. Weights bit-identical to the source builds; llama.cpp and Ollama apply the context extension with no extra flags. ## Files: MTP-first | File | Size | Pick it when | |---|---|---| | `ornith-1.0-9b-1M-MTP-Q4_K_M.gguf` | 5.8 GB | Default. MTP layer baked in: +15 to 38 percent decode via llama.cpp speculative decoding | | `ornith-1.0-9b-1M-MTP-Q8_0.gguf` | 9.8 GB | Max quality. This is the config that scored 10/10 at the full 1M with f16 KV | | `mmproj-ornith-9b-f16.gguf` | 918 MB | Vision, attach with `--mmproj` | This repo intentionally carries only the MTP builds. The **full 10-quant ladder** (IQ2_M through bf16) lives on the [ModelScope mirror](https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF). ## Every file, every mirror Nothing was discontinued: every quant is one click away. Hugging Face carries the curated picks, ModelScope always carries everything, and Ollama serves ready-to-run tags. On Ollama every tag ships with the vision tower bundled and the 1M rope metadata baked in. | File | Size | Hugging Face | ModelScope | Ollama | |---|---|---|---|---| | `mmproj-ornith-9b-f16.gguf` | 918 MB | [download](https://huggingface.co/satgeze/Ornith-1.0-9B-1M-GGUF/resolve/main/mmproj-ornith-9b-f16.gguf) | [download](https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF/resolve/master/mmproj-ornith-9b-f16.gguf) | bundled in every tag | | `ornith-1.0-9b-1M-IQ2_M.gguf` | 3.6 GB | on ModelScope | [download](https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF/resolve/master/ornith-1.0-9b-1M-IQ2_M.gguf) | [`ollama run satgeze/ornith-9b-1m:iq2_m`](https://ollama.com/satgeze/ornith-9b-1m) | | `ornith-1.0-9b-1M-IQ3_XXS.gguf` | 3.9 GB | on ModelScope | [download](https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF/resolve/master/ornith-1.0-9b-1M-IQ3_XXS.gguf) | [`ollama run satgeze/ornith-9b-1m:iq3_xxs`](https://ollama.com/satgeze/ornith-9b-1m) | | `ornith-1.0-9b-1M-IQ4_XS.gguf` | 5.2 GB | on ModelScope | [download](https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF/resolve/master/ornith-1.0-9b-1M-IQ4_XS.gguf) | [`ollama run satgeze/ornith-9b-1m:iq4_xs`](https://ollama.com/satgeze/ornith-9b-1m) | | `ornith-1.0-9b-1M-MTP-Q4_K_M.gguf` | 5.8 GB | [download](https://huggingface.co/satgeze/Ornith-1.0-9B-1M-GGUF/resolve/main/ornith-1.0-9b-1M-MTP-Q4_K_M.gguf) | [download](https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF/resolve/master/ornith-1.0-9b-1M-MTP-Q4_K_M.gguf) | [`ollama run satgeze/ornith-9b-1m`](https://ollama.com/satgeze/ornith-9b-1m) | | `ornith-1.0-9b-1M-MTP-Q8_0.gguf` | 9.8 GB | [download](https://huggingface.co/satgeze/Ornith-1.0-9B-1M-GGUF/resolve/main/ornith-1.0-9b-1M-MTP-Q8_0.gguf) | [download](https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF/resolve/master/ornith-1.0-9b-1M-MTP-Q8_0.gguf) | [`ollama run satgeze/ornith-9b-1m:q8_0`](https://ollama.com/satgeze/ornith-9b-1m) | | `ornith-1.0-9b-1M-Q2_K.gguf` | 3.8 GB | on ModelScope | [download](https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF/resolve/master/ornith-1.0-9b-1M-Q2_K.gguf) | [`ollama run satgeze/ornith-9b-1m:q2_k`](https://ollama.com/satgeze/ornith-9b-1m) | | `ornith-1.0-9b-1M-Q3_K_M.gguf` | 4.6 GB | on ModelScope | [download](https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF/resolve/master/ornith-1.0-9b-1M-Q3_K_M.gguf) | [`ollama run satgeze/ornith-9b-1m:q3_k_m`](https://ollama.com/satgeze/ornith-9b-1m) | | `ornith-1.0-9b-1M-Q4_K_M.gguf` | 5.6 GB | on ModelScope | [download](https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF/resolve/master/ornith-1.0-9b-1M-Q4_K_M.gguf) | [`ollama run satgeze/ornith-9b-1m`](https://ollama.com/satgeze/ornith-9b-1m) | | `ornith-1.0-9b-1M-Q5_K_M.gguf` | 6.5 GB | on ModelScope | [download](https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF/resolve/master/ornith-1.0-9b-1M-Q5_K_M.gguf) | [`ollama run satgeze/ornith-9b-1m:q5_k_m`](https://ollama.com/satgeze/ornith-9b-1m) | | `ornith-1.0-9b-1M-Q6_K.gguf` | 7.4 GB | on ModelScope | [download](https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF/resolve/master/ornith-1.0-9b-1M-Q6_K.gguf) | [`ollama run satgeze/ornith-9b-1m:q6_k`](https://ollama.com/satgeze/ornith-9b-1m) | | `ornith-1.0-9b-1M-Q8_0.gguf` | 9.5 GB | on ModelScope | [download](https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF/resolve/master/ornith-1.0-9b-1M-Q8_0.gguf) | [`ollama run satgeze/ornith-9b-1m:q8_0`](https://ollama.com/satgeze/ornith-9b-1m) | | `ornith-1.0-9b-1M-bf16.gguf` | 17.9 GB | on ModelScope | [download](https://www.modelscope.ai/models/satgeze/Ornith-1.0-9B-1M-GGUF/resolve/master/ornith-1.0-9b-1M-bf16.gguf) | [`ollama run satgeze/ornith-9b-1m:bf16`](https://ollama.com/satgeze/ornith-9b-1m) | ## Measured: the ladder and the quantization tax Full ladder, no skipped rungs, 10 needles per rung at depths 5 to 95 percent, temperature 0, fresh-seed replications included: ![Ornith-9B NIAH heatmap](niah_heatmap_9b.png) Two configurations tell the whole story: | Config | 64K to 524K | 1M | |---|---|---| | Budget: Q4_K_M weights + q8_0 KV (fits 24 to 32 GB GPUs) | 10/10 every rung | 7/10, misses cluster at 65 to 85 percent depth | | Max quality: Q8_0 weights + f16 KV (M3 Max 128 GB) | 10/10 | **10/10** | Same model, same 1M bake. The 1M misses in the budget config are quantization tax, not a context-extension failure; the extension itself is clean, as the max-quality run proves. Prefill time at 1M on M3 Max: about 6.8 hours (`prefill_time_9b.png`). ## MTP The Qwen3.5 family ships a multi-token-prediction layer in official checkpoints; Ornith's GGUFs dropped it and [protoLabsAI](https://huggingface.co/protoLabsAI/Ornith-1.0-9B-MTP-GGUF) restored it. Both files above carry it baked in. Measured decode gains on RTX 5090: 218 tok/s baseline to 250 to 302 tok/s depending on workload (best on code). Output is identical to standard decoding; the trunk verifies every drafted token. ```bash llama-server -m ornith-1.0-9b-1M-MTP-Q4_K_M.gguf \ -c 1048576 -np 1 --jinja \ --spec-type draft-mtp --spec-draft-n-max 3 \ --mmproj mmproj-ornith-9b-f16.gguf ``` Ollama (1M and vision work; no speculative decoding in Ollama yet): ``` FROM ./ornith-1.0-9b-1M-MTP-Q4_K_M.gguf RENDERER qwen3.5 PARSER qwen3.5 PARAMETER num_ctx 262144 ``` ## How this was built 1M context: YaRN rope-scaling metadata (factor 4.0 over native 262,144) baked into the GGUF header with gguf-py; no weight changes, no fine-tuning. MTP: community GGUF carrying the official layer, vetted and re-baked. Certification: multi-needle harness (in this repo as `niah_test.py`), f16 KV for certification runs, quantized KV only ever as a labeled budget config. Method and tooling: [github.com/satindergrewal/aviary-1m](https://github.com/satindergrewal/aviary-1m). For base capability benchmarks see the [official Ornith-1.0-9B card](https://huggingface.co/deepreinforce-ai/Ornith-1.0-9B). Sister repos: [35B](https://huggingface.co/satgeze/Ornith-1.0-35B-1M-GGUF) (50/50 through 1M + MTP), [397B](https://huggingface.co/satgeze/Ornith-1.0-397B-1M-GGUF). ## Credits Ornith-1.0: [DeepReinforce](https://huggingface.co/deepreinforce-ai) (MIT), post-trained on Qwen 3.5. MTP layer: Qwen; GGUF restoration: protoLabsAI. Vision tower: Qwen3.5 family. 1M YaRN extension, vetting, and certification: [SatGeze](https://huggingface.co/satgeze).