Instructions to use JANGQ-AI/MiMo-V2.6-Flash-MOPD-JANGHT2.5 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use JANGQ-AI/MiMo-V2.6-Flash-MOPD-JANGHT2.5 with MLX:

# Make sure mlx-vlm is installed
# pip install --upgrade mlx-vlm

from mlx_vlm import load, generate
from mlx_vlm.prompt_utils import apply_chat_template
from mlx_vlm.utils import load_config

# Load the model
model, processor = load("JANGQ-AI/MiMo-V2.6-Flash-MOPD-JANGHT2.5")
config = load_config("JANGQ-AI/MiMo-V2.6-Flash-MOPD-JANGHT2.5")

# Prepare input
image = ["http://images.cocodataset.org/val2017/000000039769.jpg"]
prompt = "Describe this image."

# Apply chat template
formatted_prompt = apply_chat_template(
    processor, config, prompt, num_images=1
)

# Generate output
output = generate(model, processor, formatted_prompt, image)
print(output)

Notebooks
Google Colab
Kaggle
Local Apps Settings
LM Studio

How to use JANGQ-AI/MiMo-V2.6-Flash-MOPD-JANGHT2.5 with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "JANGQ-AI/MiMo-V2.6-Flash-MOPD-JANGHT2.5"

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "mlx-lm": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "JANGQ-AI/MiMo-V2.6-Flash-MOPD-JANGHT2.5"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent

How to use JANGQ-AI/MiMo-V2.6-Flash-MOPD-JANGHT2.5 with Hermes Agent:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "JANGQ-AI/MiMo-V2.6-Flash-MOPD-JANGHT2.5"

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default JANGQ-AI/MiMo-V2.6-Flash-MOPD-JANGHT2.5

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use JANGQ-AI/MiMo-V2.6-Flash-MOPD-JANGHT2.5 with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "JANGQ-AI/MiMo-V2.6-Flash-MOPD-JANGHT2.5"

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "JANGQ-AI/MiMo-V2.6-Flash-MOPD-JANGHT2.5" \
  --custom-provider-id mlx-lm \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Run JANG models in MLX Studio / vMLX

⚠️ Runtime: this bundle needs a vMLX (Python) build that is not released yet. It was built and validated on a vMLX development build that adds MiMo-V2.6 JANGHT loading (JANGT trellis + JANGH codebook experts on the MiMo-V2.6 omni model). Released vMLX builds do not load it. Use it with the first vMLX release that lists MiMo-V2.6 JANGHT support.

JANGQ-AI/MiMo-V2.6-Flash-MOPD-JANGHT2.5

MiMo-V2.6-Flash-MOPD for 128 GB Macs: the full omni model (text, image, video, audio in) in 93.9 GiB, with its native DFlash drafter in the box.

A JANGHT bundle of XiaomiMiMo/MiMo-V2.6-Flash-MOPD: a 310B-parameter MoE (48 layers, 256 routed experts top-8, layer 0 dense; 9 full-attention + 39 sliding-window layers with attention sinks) with vision, video and audio input.

Routed experts: JANGHT, a per-unit mix of two formats, chosen by measurement:
- JANGT: a trellis code at 2, 2.5 or 3 bits per weight (Hadamard-128 rotation with signs, 12-bit state, up to 8 exact fp16 outlier columns per projection).
- JANGH: our Hadamard-32 codebook format at 2, 3 or 4 bits, rounded with GPTQ.
Everything else in the text model: affine 8-bit (group 64) — attention, the dense layer-0 MLP, embeddings and lm_head. Routers, expert-score bias and attention sinks stay fp32.
Vision tower, audio encoder, speech embeddings: bf16, unchanged from the source. Audio tokenizer: the encoder half, verbatim (the decoder only produces speech and is not used for input).
DFlash drafter (dflash/): MiMo's native block-diffusion draft model (MIT), 2-D weights affine 8-bit.
No MTP: the three MTP layers are omitted (num_nextn_predict_layers: 0); DFlash replaces them.
Name: JANGHT + 2.5 bits per weight averaged over the whole model (91.84 GiB of model shards × 8 / 310.76B source parameters = 2.539), drafter and audio tokenizer excluded.

Quality vs the original model

254 held-out sequences, 153,581 teacher-forced positions, scored against the original checkpoint (official MXFP4 experts decoded exactly, FP8 attention dequantized per block, bf16 elsewhere) run layer by layer from disk. Top-128 renormalized KL. Most positions are prompts and conversations from public datasets (multi-turn tool use, image and audio questions, long documents); every assistant turn was written by MiMo-V2.6-Flash-MOPD itself (T=1.0, top-p 0.95, thinking on and off), so on those turns agreement measures how often the quant makes the same next-token choice as the original model on its own text.

	experts	non-experts	median KL ↓	mean KL ↓	p90 / p95 / p99 ↓	top-1 ↑	top-5 ↑	top-10 ↑
harness check: original vs its own reference (68,811 positions)	original	original	0.0000	0.0000	0.00 / 0.00 / 0.00	100.0%	100.0%	100.0%
floor: original experts, 8-bit non-experts	original	affine 8-bit	0.079	0.931	3.01 / 5.69 / 10.40	69.6%	84.6%	87.7%
floor: original experts, MXFP8 non-experts	original	MXFP8	0.100	1.036	3.48 / 6.16 / 10.73	67.9%	83.1%	86.3%
MiMo-V2.6-Flash-MOPD-JANGHT2.5	JANGHT	affine 8-bit	0.519	1.758	5.51 / 7.86 / 12.06	54.8%	71.7%	76.5%

How to read this:

MiMo is sensitive to tiny changes on long, high-entropy text. Rounding only the non-expert weights to 8 bits, with the experts untouched, already drops top-1 agreement to 69.6% overall and to 58–67% on image-conversation and tool-conversation text. No quant of this model reaches 100% on those sets; the floor rows are the realistic ceiling. Affine 8-bit beat MXFP8 on every domain and is what ships.
On reasoning, chat and security text the bundle stays close: 91–93% top-1 agreement, median KL 0.001–0.007.
Next-token accuracy on the actual held-out text: 34.9% for the bundle vs 36.0% for the original model.

By domain, median KL · mean KL · top-1 agreement:

domain	positions	JANGHT2.5	floor (original experts, 8-bit non-experts)
tool conversations (multi-turn, public datasets)	84,770	0.941 · 1.861 · 48.7%	0.145 · 0.809 · 67.2%
image conversations	34,192	0.827 · 2.612 · 46.3%	0.219 · 1.810 · 58.0%
prose / long documents	11,650	0.192 · 1.037 · 67.1%	0.036 · 0.517 · 80.0%
coding	10,604	0.027 · 0.715 · 79.9%	0.003 · 0.515 · 85.7%
reasoning	3,483	0.001 · 0.075 · 93.5%	0.000 · 0.014 · 97.9%
cybersecurity	2,894	0.002 · 0.066 · 93.0%	0.000 · 0.013 · 97.9%
Chinese	2,013	0.031 · 0.079 · 88.6%	0.002 · 0.006 · 97.0%
agentic	1,401	0.362 · 1.088 · 61.4%	0.029 · 0.387 · 78.7%
audio conversations	1,353	0.091 · 0.777 · 74.0%	0.004 · 0.262 · 89.2%
chat	1,221	0.007 · 0.111 · 91.0%	0.000 · 0.023 · 97.3%

Tool decisions

Inside the 153k positions: 89 points where the original model's own transcript starts a tool call, 149 where it starts the answer (first token after </think>).

	JANGHT2.5	floor (8-bit non-experts)
tool-call points: same next token as the original	98.9%	100%
tool-call points: model starts a tool call (original: 89 of 89)	88	89
median P(`<tool_call>`) at tool-call points (original: 0.9994)	0.996	0.999
lowest P(`<tool_call>`) at a tool-call point	0.28	0.72
answer points: same next token as the original	73.2%	94.6%
decisions flipped call ↔ answer (all points)	3	0
tool-call tokens: next-token accuracy (original: 98.8%)	97.6%	98.8%

Reasoning ends on its own

At the 231 points where the original model closes its reasoning (</think>) in its own transcripts:

	JANGHT2.5	floor (8-bit non-experts)
median P(`</think>`) (original: 0.993)	0.964	0.989
points where P(`</think>`) < 0.1	7 of 231	2 of 231

Every expert unit carries the unit-gain row scales (the fix for low-bit reasoning that never closes, see our GLM-5.3-Flash JANGH2 revision 2).

Free generation (served, thinking on, T=1.0, top-p 0.95, 4 hard prompts × 3 seeds, 12k-token cap): 11 / 12 runs closed their reasoning by themselves with no repetition, after 268 to 9,003 characters of reasoning. The twelfth (the 23-person birthday problem, seed 2) was still multiplying the 22 factors out by hand, digit by digit and without repeating itself, when it reached the 12k-token cap; the other two seeds of that prompt closed. Re-run alone with a 32k-token cap, that prompt and seed still had not closed after 32,000 reasoning tokens. Long hand-arithmetic derivations can overrun the reasoning budget; set a generous max_tokens.

Live behavior (served, vMLX, plain lane)

suite	JANGHT2.5, omni route (default)	JANGHT2.5, text-only route (`--text-only`)
behavior probes: math (thinking on / off), follow-up, 8 tool calls (required / auto × thinking on / off), tool round trip, prime check (thinking on / off), image, video, audio transcription	16 / 16	13 / 13 (no media)
48-case tool-use eval (required / auto × thinking on / off), temperature 0	46 / 48	46 / 48
multi-round agent episodes vs a scripted environment (8 tasks × thinking on / off × 2 seeds, T=1.0): no repeated tool calls, task completed	30 / 32	—

The two tool-use misses are the same case: asked "how many kilometers is 26.2 miles?" with thinking on, the model called calculator with 26.2 * 1.609344 instead of convert_units.
Agent episodes: 160 tool calls, 1 repeat (page 2 of a search fetched twice, then a correct summary). One episode ignored a tool error that said the exchange was closed and answered that the ticker was not real. The test-fix loop (run tests → read → edit → re-run until green), schema exploration and a 5-item to-do were completed in every run.
Audio transcription of a spoken sentence: "The quick brown fox jumps over the lazy dog near the river bank." — exact.

Speed (M5 Max 128 GB)

Served, one request at a time, plain lane, median of 3 probes on never-seen prompts (one server run).

	JANGHT2.5
decode (tok/s)	45.9 (45.9 – 46.1)
prefill, ~5.1k-token prompt (tok/s)	692
time to first token, ~5.1k-token prompt	7.4 s
Metal memory after load	91.5 GB active (~100 GB peak during an image request)

All numbers on this card are from the plain lane. vMLX support for the bundled DFlash drafter is in progress; until it ships, serve with --no-bundled-dflash2.

What's in the bundle

Experts (47 MoE layers; a unit is the gate+up pair or the down projection of one layer):

projection	JANGT 2	JANGT 2.5	JANGT 3	JANGH 2	JANGH 3	JANGH 4
gate / up	20 layers	22	2	2	—	1
down	8	24	—	12	1	2

Layers 5 and 16 are all-JANGH; the per-layer map is in config.json → quantization and jang_config.json → expert_units.

Per-layer formats (measured from the shards)

Layer 0 is the dense MLP (affine 8-bit). Type = attention (full, or sliding window 128 with sinks). Outlier = exact fp16 columns per JANGT projection.

layer	type	gate	up	down	outlier g/u/d	GiB
1	SWA-128	JANGT2	JANGT2	JANGT2	8/8/8	1.557
2	SWA-128	JANGT2	JANGT2	JANGT2	8/8/8	1.557
3	SWA-128	JANGT2	JANGT2	JANGT2	8/8/8	1.557
4	SWA-128	JANGT2	JANGT2	JANGT2	8/8/8	1.557
5	full	JANGH2	JANGH2	JANGH2	–/–/–	1.504
6	SWA-128	JANGH2	JANGH2	JANGT2	–/–/8	1.531
7	SWA-128	JANGT2	JANGT2	JANGT2	8/8/8	1.557
8	SWA-128	JANGT2	JANGT2	JANGT2.5	8/8/8	1.682
9	SWA-128	JANGT2	JANGT2	JANGH2	5/5/–	1.523
10	SWA-128	JANGT3	JANGT3	JANGH4	3/3/–	2.520
11	full	JANGT2.5	JANGT2.5	JANGH3	2/2/–	2.018
12	SWA-128	JANGT2	JANGT2	JANGH2	1/1/–	1.516
13	SWA-128	JANGT2	JANGT2	JANGT2.5	0/0/8	1.668
14	SWA-128	JANGT2	JANGT2	JANGT2.5	0/0/8	1.668
15	SWA-128	JANGT2	JANGT2	JANGH2	0/0/–	1.516
16	SWA-128	JANGH4	JANGH4	JANGH4	–/–/–	3.004
17	full	JANGT2	JANGT2	JANGT2.5	0/0/8	1.668
18	SWA-128	JANGT2	JANGT2	JANGH2	0/0/–	1.516
19	SWA-128	JANGT2	JANGT2	JANGT2.5	0/0/8	1.668
20	SWA-128	JANGT2.5	JANGT2.5	JANGT2.5	0/0/8	1.918
21	SWA-128	JANGT2.5	JANGT2.5	JANGT2.5	0/0/8	1.918
22	SWA-128	JANGT2.5	JANGT2.5	JANGT2.5	1/1/8	1.918
23	full	JANGT2.5	JANGT2.5	JANGT2.5	1/1/8	1.918
24	SWA-128	JANGT2.5	JANGT2.5	JANGT2.5	1/1/8	1.918
25	SWA-128	JANGT2.5	JANGT2.5	JANGT2.5	1/1/8	1.918
26	SWA-128	JANGT2.5	JANGT2.5	JANGT2.5	1/1/8	1.918
27	SWA-128	JANGT2.5	JANGT2.5	JANGT2.5	1/1/8	1.918
28	SWA-128	JANGT2.5	JANGT2.5	JANGT2	2/2/8	1.795
29	full	JANGT2.5	JANGT2.5	JANGT2.5	2/2/8	1.920
30	SWA-128	JANGT2	JANGT2	JANGH2	2/2/–	1.518
31	SWA-128	JANGT2.5	JANGT2.5	JANGT2.5	2/2/8	1.920
32	SWA-128	JANGT2	JANGT2	JANGH2	1/1/–	1.516
33	SWA-128	JANGT2	JANGT2	JANGH2	2/2/–	1.518
34	SWA-128	JANGT2	JANGT2	JANGH2	1/1/–	1.516
35	full	JANGT2	JANGT2	JANGH2	2/2/–	1.518
36	SWA-128	JANGT2.5	JANGT2.5	JANGT2.5	0/0/8	1.918
37	SWA-128	JANGT2	JANGT2	JANGH2	1/1/–	1.516
38	SWA-128	JANGT2.5	JANGT2.5	JANGT2.5	0/0/8	1.918
39	SWA-128	JANGT2.5	JANGT2.5	JANGT2.5	1/1/8	1.918
40	SWA-128	JANGT2.5	JANGT2.5	JANGT2.5	2/2/8	1.920
41	full	JANGT3	JANGT3	JANGT2	2/2/8	2.045
42	SWA-128	JANGT2.5	JANGT2.5	JANGT2.5	2/2/8	1.920
43	SWA-128	JANGT2.5	JANGT2.5	JANGT2.5	2/2/8	1.920
44	SWA-128	JANGT2.5	JANGT2.5	JANGT2.5	2/2/8	1.920
45	SWA-128	JANGT2.5	JANGT2.5	JANGT2.5	1/1/8	1.918
46	SWA-128	JANGT2.5	JANGT2.5	JANGT2.5	1/1/8	1.918
47	full	JANGT2.5	JANGT2.5	JANGH2	1/1/–	1.766

Model shards: 18 shards, 91.84 GiB. Routed experts 83.94 GiB; text non-experts affine 8-bit; vision tower, audio encoder and speech embeddings bf16.
audio_tokenizer/ (0.61 GiB): encoder only, verbatim.
dflash/ (1.45 GiB): DFlash drafter — 5 sliding-window (1024) layers reading the target's hidden states from layers 0 / 11 / 23 / 35 / 47, block size 8, sharing the target's embeddings and lm_head; 36 linears affine 8-bit g64, mask_embedding.safetensors (+ the original .pt), MIT.
Thinking + agentic:
- Thinking is ON by default; pass chat_template_kwargs: {"enable_thinking": false} to turn it off. There are no reasoning-effort levels.
- Tool calls use MiMo's XML dialect (<tool_call><function=name><parameter=key>value</parameter></function></tool_call>), declared as tool_parser: xml_function; reasoning as think_xml.
- Past assistant turns keep their reasoning_content in the template; send it back in multi-turn agent loops.
Self-describing:
- config.json carries the per-module quantization map: 120 JANGT expert projections ("mode": "jangt", "kbits"), 21 JANGH ("mode": "jangtq2", "bits", "rotation": "hadamard32"), affine 8-bit (group 64) for every other quantized linear. The jangt block holds the trellis parameters and the per-projection outlier columns; the jangtq block (version 2) holds the JANGH codebooks.
- jang_config.json records the calibration, the allocation plan, the per-unit formats, the drafter and the capabilities.
Every tensor file is alignment-safe (zero-copy memory mapping), including the drafter and audio tokenizer.

Serving contract

Sampling: temperature=1.0, top_p=0.95 (model card defaults), no repetition penalty
EOS: [151643, 151645, 151672] · context: 1M declared (memory, not the model, sets the practical limit on 128 GB)
Thinking: enable_thinking chat-template kwarg, default on
JANGH units keep their on-disk identifiers (jangtq block, "mode": "jangtq2", *.tq2_packed / *.tq2_scales). They are not loadable by, and must not be routed to, JANGTQ v1 loaders.

Build details

Source: XiaomiMiMo/MiMo-V2.6-Flash-MOPD @ b12f557; experts decoded exactly from the official MXFP4 (QAT) weights
Calibration: 1.29M tokens. Every assistant turn was generated by the source model itself (364k generated tokens, thinking on and off), over coding in 9 languages, multi-round agentic tool use, tool conversations, terminal / shell, cybersecurity (two public instruction and vulnerability-fix datasets), reasoning, chat, Chinese, prose, 165 images and 60 audio clips run through the bf16 towers. Held out: 254 sequences (154k tokens) never used for calibration
Allocation: per unit (gate+up or down of each layer), the format and width with the lowest predicted cost: measured held-out error curves weighted by each layer's measured sensitivity (real quantization error, frozen routing)
Rounding: JANGT trellis on per-expert input statistics with outlier columns and column scaling on down projections; JANGH GPTQ with act-order in the rotated basis; unit-gain row scales on every unit
Drafter: MiMo's native DFlash drafter, weights rounded to affine 8-bit; architecture unchanged

Quantized and validated by Jinho Jang — eric@jangq.ai

Downloads last month: 153

Safetensors

Model size

30B params

Tensor type

F32

I32

U32

BF16

F16

MLX

Hardware compatibility

Quantized

Model tree for JANGQ-AI/MiMo-V2.6-Flash-MOPD-JANGHT2.5

Base model

XiaomiMiMo/MiMo-V2.6-Flash-MOPD

Quantized

(30)

this model