GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
88plug
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
88plug
Model Tree
Input Modalities
Output Modalities
Supported Functionality
Intended format is compressed-tensors (pack-quantized). Current Hub files fail pack validation — see warning above.
| Runtime | Supported |
|---|---|
| vLLM ≥ 0.21 | Yes — preferred (auto-detect CT; no --quantization flag) |
transformers + compressed-tensors | Yes for many text models; multimodal may need custom code |
| Text Generation Inference (TGI) | Not supported for these CT packs |
| Hugging Face Inference Widget | Often fails — use vLLM locally instead |
bash
# Preferredvllm serve 88plug/<ModelName> --trust-remote-code
Do not deploy via TGI “text-generation-inference” paths — that backend does not load our CT format and produces opaque worker/load errors.
INT4 post-training quantization of openbmb/MiniCPM-o-4.5 — a compact omni model with vision (SigLIP2), audio (Whisper), and speech synthesis (CosyVoice2) built on a Qwen3-8B backbone. ~4–5 GB on disk. Runs on a single 8 GB GPU.
| Property | Value |
|---|---|
| Base model | openbmb/MiniCPM-o-4.5 |
| Release tier | Pending-gold (Hub pack incomplete — re-quant queued) |
| Quant method | AutoRound W4A16 target; current Hub dump incomplete (no weight_packed) |
| FLAC status | Not measured (T+7d milestone) |
| Architecture | Qwen3-8B LLM + SigLIP2 vision + Whisper audio + CosyVoice2 TTS |
| Quant format | compressed-tensors (native vLLM) |
| Scheme | W4A16 |
| Group size |
| Configuration | BF16 | W8A16 | W4A16 |
|---|---|---|---|
| Weights | ~18 GB | ~9 GB | ~4–5 GB |
| Min GPU | 1× A100 40 GB | 1× RTX 3090 24 GB | 1× RTX 3080 10 GB |
Note: The non-quantized modal encoders (SigLIP2 ~1 GB, Whisper ~390 MB, CosyVoice2 ~100 MB) are included in all footprint estimates above. Only the Qwen3-8B LLM backbone is quantized to 4-bit.
Tested with vLLM v0.21.0 (vllm/vllm-openai:v0.21.0-cu129-ubuntu2404). Weights are in compressed-tensors format — vLLM detects and loads quantization automatically. No --quantization flag needed.
bash
docker run --gpus device=0 -p 8080:8080 \vllm/vllm-openai:v0.21.0-cu129-ubuntu2404 vllm serve \88plug/MiniCPM-o-4.5-W4A16 \--kv-cache-dtype fp8 \--max-model-len 32768 \--gpu-memory-utilization 0.90
Weights are in compressed-tensors format — no --quantization flag needed. Mainline vLLM returns text only; CosyVoice2 TTS output is not supported.
python
from openai import OpenAIclient = OpenAI(base_url="http://localhost:8080/v1", api_key="token")response = client.chat.completions.create(model="88plug/MiniCPM-o-4.5-W4A16",messages=[{"role": "user", "content": "Describe the architecture of MiniCPM-o 4.5."}],max_tokens=512,)print(response.choices[0].message.content)
Only the Qwen3-8B LLM backbone (model.llm) is quantized. AutoRound (blocked — see KNOWN-FAILURES) applies W4A16 to all Linear layers within model.llm, using a round-to-nearest rotation-based optimization with 200 calibration iterations per block.
| Component | Module path | Precision | Reason |
|---|---|---|---|
| Vision encoder | vision_model.* | BF16 | Excluded from recipe |
| Audio encoder | audio_model.* | BF16 | Excluded from recipe |
| CosyVoice2 TTS decoder | tts.* | BF16 | Excluded from recipe |
| Embedding layers |
The full MiniCPM-o-4.5 checkpoint is saved via model.save_pretrained() after in-place quantization of model.llm, so the output contains all modalities — vision, audio, and TTS encoders remain at full BF16 fidelity.
MiniCPM-o-4.5 required four patches to run cleanly through llmcompressor:
get_imports patch — filters minicpmo, librosa, and soundfile imports to avoid the librosa→soxr cascade during quantization.MiniCPMTTSConfig.__getattr__ patch — backfills top_p, top_k, and related attributes missing from the shipped config.json._move_missing_keys_from_meta_to_device wrap — handles all_tied_weights_keys not being set by MiniCPMO's remote code under transformers 5.8.1.is_mllm_model=False override — forces AutoRound (blocked — see KNOWN-FAILURES) through the standard LLM path instead of the multimodal MLLM compressor, which would fail trying to load a processor from directly.Additionally, torch.nn.Module.apply and torch.nn.Module.train are replaced with iterative equivalents to avoid stack overflow on MiniCPM-o's ~985-deep module tree.
| Metric | Target |
|---|---|
| KL divergence vs BF16 | < 0.014 |
| MMLU recovery | ≥ 99% |
| RULER@128k | ≥ 97% |
MiniCPM-o-4.5 is an omni model — meaningful comparisons must also support vision + audio input. As of publication, no other compressed-tensors or vLLM-native quantization of this model exists.
| Model | Source | Format | Compare angle |
|---|---|---|---|
openbmb/MiniCPM-o-4.5 | official | BF16 | Quality ceiling |
88plug/MiniCPM-o-4.5-W8A16 | 88plug | compressed-tensors W8A16 | Higher-precision sibling |
88plug/MiniCPM-o-4.5-W4A16 | 88plug | compressed-tensors W4A16 | This model |
First-to-market claim: No compressed-tensors or vLLM-native W4A16 quant was found for this model at publication time. This is the only production-ready W4A16 quant for direct vLLM serving.
Results pending.
| Engine | Format | Batch | ctx | tok/s | TTFT p50 | TTFT p99 | VRAM |
|---|---|---|---|---|---|---|---|
| vLLM v0.21.0 | W4A16 compressed-tensors | 1 | 32k | — | — | — | — |
| vLLM v0.21.0 | W4A16 compressed-tensors | 8 | 32k | — |
Hardware: A6000 48 GB, CUDA 12.9, driver 570.
SGLang does not natively support compressed-tensors. To use this model with SGLang, serve the BF16 base (openbmb/MiniCPM-o-4.5) or an AWQ variant.
bash
docker run --gpus device=0 -p 30000:30000 \lmsysorg/sglang:v0.5.8-cu129 python -m sglang.launch_server \--model-path openbmb/MiniCPM-o-4.5 \--tp 1 \--mem-fraction-static 0.85 \--port 30000
SGLang results are BF16 baseline — useful as a throughput ceiling reference, not a direct quality comparison to this quant.
Mainline llama.cpp supports MiniCPM-V (vision + text). For full CosyVoice2 speech output, use the tc-mb/llama.cpp-omni fork. Convert and quantize from the BF16 base — do not convert from compressed-tensors weights.
bash
python convert_hf_to_gguf.py openbmb/MiniCPM-o-4.5 \--outfile MiniCPM-o-4.5-BF16.ggufllama-quantize MiniCPM-o-4.5-BF16.gguf MiniCPM-o-4.5-Q4_K_M.gguf Q4_K_Mllama-quantize --imatrix calibration_datav3.txt \MiniCPM-o-4.5-BF16.gguf MiniCPM-o-4.5-IQ4_XS.gguf IQ4_XSllama-server \--model MiniCPM-o-4.5-Q4_K_M.gguf \--n-gpu-layers 999 \--ctx-size 32768 \--port 8081
| Parameter | Value |
|---|---|
| Quantizer | AutoRound (blocked — see KNOWN-FAILURES) (via llmcompressor AutoRound (blocked — see KNOWN-FAILURES)Modifier) |
| Targets | ["Linear"] within model.llm |
| Scheme | W4A16 |
| Pipeline | sequential |
| Max seq length | 2048 |
| Ignore list | , , |
bibtex
@misc{minicpmo,title = {MiniCPM-o: A GPT-4o Level Multimodal LLM on Your Phone},author = {MiniCPM Team, OpenBMB},year = {2025},url = {https://huggingface.co/openbmb/MiniCPM-o-4.5}}
88plug AI Lab ships FLAC-target compressed-tensors quantizations — AutoRound iters=200, native vLLM v0.21.0+, no extra flags.
This release: Gold tier — full AutoRound calibration (1024 samples, UltraChat + WikiText-103). Targets ≥99% MMLU recovery.
All weights use compressed-tensors format. vLLM reads quantization_config automatically.
Browse all releases → huggingface.co/88plug
| default (128) |
| Quantized | model.llm transformer Linear layers (Qwen3-8B backbone) |
| Kept BF16 | Vision encoder (SigLIP2), audio encoder (Whisper), TTS (CosyVoice2), embeddings, LM head, norms |
| Disk size | ~19 GB on Hub (unpacked — true W4 target ~4–5 GB after re-quant) |
| Min GPU | 1× RTX 3080 10 GB |
re:.*embed_tokens$ |
| BF16 |
| Standard practice (ignore list) |
| Layer norms | re:.*norm$ | BF16 | Standard practice (ignore list) |
| LM head | lm_head | BF16 | Standard practice (ignore list) |
model.llm| — |
| — |
| — |
| vLLM v0.21.0 | W4A16 compressed-tensors | 1 | 128k | — | — | — | — |
| SGLang v0.5.8 | BF16 (baseline) | 1 | 32k | — | — | — | — |
| llama.cpp b9297 | Q4_K_M GGUF | 1 | 32k | — | — | — | — |
| llama.cpp b9297 | IQ4_XS GGUF | 1 | 32k | — | — | — | — |
lm_headre:.*embed_tokens$re:.*norm$| Activations | FP16 (unquantized — W4A16) |
| trust_remote_code | required |