CMS Manhattan is a member of the NVIDIA Inception Program.
A 180B MoE model built on the Qwen3.8-Flash-Next architecture: Gated DeltaNet linear attention interleaved with Qwen Sparse Attention (QSA), 512 experts with top-10 routing plus a shared expert, gated residual (hyper-connection) streams, and a 51B-parameter hashed n-gram embedding table. Only ~6B parameters are computed per token, so throughput on CPU is closer to a 6B dense model while quality is that of a much larger one.
| Stage | Status |
|---|---|
Hand-rolled JiRack port of the architecture (JiRackDeltaNet_180b.py) |
✅ done |
| Conversion from Qwen/Qwen3.8-Flash-Next and numerical verification against the reference | ✅ done — next-token prediction matches the reference on 100% of positions; logit differences are within the model's own bf16 noise floor |
| BitLinear patch (74,028 layers) at lambda = 0 | ✅ done — bit-exact passthrough (max logit difference 0.000000) |
| Export to Hugging Face layout and llama.cpp GGUF (BF16, Q4_K_M, Q3_K_M) | ✅ done — export verified key-for-key against the original; GGUF files published (see GGUF files) |
| Ternary QAT (lambda warmup 0 → 1) | ⏳ not started — requires a machine with ≥ 640 GB RAM |
| TQ2_0 GGUF (after QAT), Ollama and Docker builds | ⏳ not yet published |
Important: until QAT is complete, the JiRack 180B weights are numerically identical to the base model (lambda = 0 means BitLinear is a pure passthrough). The ternary benefits described on this card apply after QAT.
qwen4exp architecture.<think>...</think>); JiRack builds will ship with reasoning disabled by default, as for JiRack 27B.Published in this repository. Before QAT all of them are quantizations of the unchanged base-model weights.
| File | Size | RAM (file + context) | Description |
|---|---|---|---|
JiRackDeltaNet_180b-00001-of-00007.gguf … -00007-of-00007.gguf |
354.1 GB (7 parts) | 384 GB+ | BF16, full-precision reference; load the first part, llama.cpp reads the rest |
JiRackDeltaNet_180b_Q4_K_M.gguf |
119.2 GB | 160–192 GB | Recommended balance |
JiRackDeltaNet_180b_Q3_K_M.gguf |
94.1 GB | 128 GB | Smaller, good quality / size trade-off |
Planned: TQ2_0 after QAT (~65–95 GB, estimate; depends on how the 51B n-gram table is stored).
Needs a llama.cpp build with the qwen4exp architecture (the one used to make these files).
llama-server -m JiRackDeltaNet_180b_Q4_K_M.gguf -c 8192 --jinja --reasoning-budget 0 --port 8080
--reasoning-budget 0 turns the base model's default thinking off. The server is OpenAI-compatible (/v1/chat/completions).
The JiRack checkpoint is a directory of safetensors shards that is memory-mapped lazily, so it runs even when the model is larger than RAM (slower: pages are read from disk as experts are used).
python chat_jirack_180b.py # interactive chat
The GGUF files listed above are ready to use. For other quantizations (or TQ2_0 after QAT), contact us.
These commands will work once the 180B builds are published at https://ollama.com/cmsmanhattan; until then use the GGUF files with llama.cpp (above).
ollama pull JiRackDeltaNet_180b-q4
ollama run JiRackDeltaNet_180b-q4 "What is the capital of France?"
From the base model card:
temperature=1.0, top_p=0.95, top_k=20, min_p=0.0, presence_penalty=0.0, repetition_penalty=1.0temperature=0.7, top_p=0.80, top_k=20, min_p=0.0, presence_penalty=1.5, repetition_penalty=1.0Once a JiRack container is running, open your browser at http://localhost:7869. This opens the JiRack UI — a clean web interface. The port can be changed from the Settings panel.
This repository is dual-licensed: the model weights and the JiRack code are licensed separately.
| Component | License |
|---|---|
| Model weights (safetensors, GGUF) — a derivative of Qwen/Qwen3.8-Flash-Next | Qwen Community License 1.0 — LICENSE, Copyright 2026 Qwen |
JiRack code and tools (JiRackDeltaNet_180b.py, chat_jirack_180b.py) |
MIT License — © 2025-2026 Konstantin Vladimirovich Grabko, CMS Manhattan JiRack Technology |
| JiRack Docker images with UI, pre-built Ollama quantizations, JiRack UI clients | Commercial license (see below) |
For commercial licensing, cluster deployment, or enterprise use of JiRack models, please contact us.
| Use Case | CPU | RAM | Recommended Quant | Notes |
|---|---|---|---|---|
| Recommended | High-core server CPU (Xeon / EPYC) | 160–192 GB | Q4_K_M (119.2 GB), or TQ2_0 after QAT | Only ~6B active parameters per token |
| High Performance | Dual-socket server | 384 GB+ | BF16 (354.1 GB) | Reference quality |
| Low Memory | Modern 16+ core CPU + fast NVMe | 128 GB | Q3_K_M (94.1 GB) | Usable |
| PyTorch checkpoint | Server CPU + NVMe | 256 GB+ | BF16 (lazy mmap) | Works below model size, slower |
qwen4_exp in Hugging Face, qwen4exp in llama.cpp)in_proj_qkv / in_proj_z / out_proj, attention q/k/v/o_proj, all routed experts and the shared expert — 74,028 layersA_log / dt_bias, QSA indexer, PLE projections and n-gram table, embeddings, lm_head, MTP head{-γ, 0, +γ}; activations during training: per-token int8JiRack DeltaNet 180B is built on Qwen3.8-Flash-Next. The tables below reproduce the published base-model results from Qwen/Qwen3.8-Flash-Next for reference — they reflect the upstream base model's capabilities, not JiRack-specific QAT or quantization results.
| Qwen3.8-Flash-Next | Qwen3.8-27B | Qwen3.7-Plus | DeepSeek-V4-Flash-0731 | Claude-Opus-4.6 (Max) | |
|---|---|---|---|---|---|
| # Params | 125B | 27B | 397B | 284B | -- |
| # Activated params | 6B | 27B | 17B | 13B | -- |
| # N-gram embedding params | 51B | -- | -- | -- | -- |
| Coding | |||||
| Agentic coding — DeepSWE 1.1 | 58.7 | 42.2 | 16.5 | 54.4 | -- |
| Agentic coding — SWE-bench Pro | 62.5 | 61.7 | 55.8 | 56.0 | 53.4 |
| Multilingual software engineering — SWE-bench Multilingual | 81.0 | 73.8 | 75.8 | -- | 77.5 |
| Repo-level code generation — NL2Repo-Bench | 48.1 | 42.3 | 41.1 | 54.2 | 47.6 |
| Agent | |||||
| Long-horizon office work — CoWorkBench | 73.9 | 70.7 | 65.1 | 45.1 | 68.2 |
| Professional job tasks — JobBench | 55.7 | 33.4 | 27.6 | 41.3 | 36.6 |
| Frontier agentic tasks — Agents' Last Exam (Pass@1 / Score) | 24.3 / 51.2 | 20.4 / 42.9 | 13.2 / 33.6 | 25.2 / -- | -- |
| Real-world tool use — Toolathlon Verified (Pass@1) | 73.5 | 67.1 | 50.6 | 70.3 | -- |
| General | |||||
| Instruction following — IFBench | 81.3 | 79.5 | 79.1 | 79.2 | 62.5 |
| Scientific reasoning — GPQA Diamond | 91.7 | 89.2 | 90.3 | 90.8 | 91.3 |
| Multidisciplinary reasoning — HLE | 35.9 | 30.8 | 34.7 | 33.8 | 40.0 |
| Competitive coding — LiveCodeBench v6 | 91.9 | 90.3 | 89.6 | 90.6 | 88.8 |
| Qwen3.8-Flash-Next | Qwen3.8-27B | Qwen3.7-Plus | Claude-Opus-4.6 (Max) | |
|---|---|---|---|---|
| Agentic Multimodal Intelligence | ||||
| Multimodal tool use — ClawEval-MM (Pass@3 / Average) | 64.4 / 60.4 | 57.4 / 56.9 | 57.4 / 60.1 | 52.5 / 54.7 |
| Application recreation — RecreationBench | 49.9 | 47.1 | 30.2 | -- |
| Mobile use — AndroidWorld | 84.5 | 81.9 | 81.0 | 62.0 |
| Computer use — OSWorld 2.0 (Binary / Partial) | 19.4 / 52.3 | 19.4 / 48.0 | 2.8 / 21.5 | -- |
| Visual web development — Vision2Web | 64.0 | 62.9 | 42.1 | -- |
| General Multimodal Intelligence | ||||
| Embodied intelligence — ERQA | 72.3 | 65.5 | 69.8 | 40.8 |
| Long video understanding — LVBench | 76.6 | 72.4 | 76.2 | 63.0 |
| Real-world perception — RealWorldQA | 88.5 | 85.9 | 86.9 | 73.9 |
| Visual math — MathVision (w/o CI / w/ CI) | 90.6 / 95.7 | 90.0 / 94.6 | 90.3 / 88.7 | 65.5 / -- |
| Scientific chart analysis — CharXiv (RQ) (w/o CI / w/ CI) | 84.6 / 90.6 | 83.7 / 90.2 | 85.8 / 85.9 | 66.0 / -- |
Source: Qwen/Qwen3.8-Flash-Next model card. Best result in each row is bolded. Empty cells (--) indicate results not available or not applicable. See the source card for evaluation harnesses, settings and footnotes.
@techreport{qwen2026design,
title = {On the Design of {Qwen3.8-Next} Architecture: Evaluation, Efficiency, and Training Stability},
author = {{Qwen Team}},
institution = {Alibaba Group},
month = {August},
year = {2026}
}
For joint venture opportunities, hardware integration, or licensing inquiries:
LICENSE-CODE. © 2025-2026 Konstantin Vladimirovich Grabko, CMS Manhattan JiRack Technology.JiRackDeltaNet_180b-00002-of-00007.gguf 102 GBJiRackDeltaNet_180b-00005-of-00007.gguf 49 GBJiRackDeltaNet_180b-00003-of-00007.gguf 49 GBJiRackDeltaNet_180b-00004-of-00007.gguf 48 GBJiRackDeltaNet_180b-00006-of-00007.gguf 48 GBJiRackDeltaNet_180b-00007-of-00007.gguf 46 GBJiRackDeltaNet_180b-00001-of-00007.gguf 12 GBcheckpoint-00001-of-00043.safetensors 8.6 GBcheckpoint-00025-of-00043.safetensors 8.6 GBcheckpoint-00037-of-00043.safetensors 8.6 GBcheckpoint-00022-of-00043.safetensors 8.6 GBcheckpoint-00028-of-00043.safetensors 8.6 GBcheckpoint-00031-of-00043.safetensors 8.6 GBcheckpoint-00034-of-00043.safetensors 8.6 GBcheckpoint-00040-of-00043.safetensors 8.6 GBcheckpoint-00016-of-00043.safetensors 8.6 GBcheckpoint-00019-of-00043.safetensors 8.6 GBcheckpoint-00015-of-00043.safetensors 8.6 GBcheckpoint-00039-of-00043.safetensors 8.6 GBcheckpoint-00032-of-00043.safetensors 8.6 GBcheckpoint-00023-of-00043.safetensors 8.6 GBcheckpoint-00030-of-00043.safetensors 8.6 GBcheckpoint-00027-of-00043.safetensors 8.6 GBcheckpoint-00020-of-00043.safetensors 8.6 GBcheckpoint-00035-of-00043.safetensors 8.6 GBcheckpoint-00018-of-00043.safetensors 8.6 GBcheckpoint-00036-of-00043.safetensors 8.6 GBcheckpoint-00033-of-00043.safetensors 8.6 GBcheckpoint-00026-of-00043.safetensors 8.6 GBcheckpoint-00029-of-00043.safetensors 8.6 GBcheckpoint-00021-of-00043.safetensors 8.6 GBcheckpoint-00024-of-00043.safetensors 8.6 GBcheckpoint-00038-of-00043.safetensors 8.6 GBcheckpoint-00041-of-00043.safetensors 8.6 GBcheckpoint-00017-of-00043.safetensors 8.6 GBcheckpoint-00042-of-00043.safetensors 8.3 GBcheckpoint-00013-of-00043.safetensors 8.0 GBcheckpoint-00014-of-00043.safetensors 8.0 GBcheckpoint-00012-of-00043.safetensors 8.0 GBcheckpoint-00003-of-00043.safetensors 8.0 GBcheckpoint-00004-of-00043.safetensors 8.0 GBcheckpoint-00005-of-00043.safetensors 8.0 GBcheckpoint-00006-of-00043.safetensors 8.0 GBcheckpoint-00007-of-00043.safetensors 8.0 GBcheckpoint-00008-of-00043.safetensors 8.0 GBcheckpoint-00009-of-00043.safetensors 8.0 GBcheckpoint-00010-of-00043.safetensors 8.0 GBcheckpoint-00011-of-00043.safetensors 8.0 GBcheckpoint-00002-of-00043.safetensors 7.8 GBcheckpoint-00043-of-00043.safetensors 6.5 GB