MiniMax-M2.5-quantized.w8a8

Model Overview

• Model Architecture: MiniMaxM2ForCausalLM
  • Input: Text
  • Output: Text
• Model Optimizations:
  • Weight quantization: INT8
• Intended Use Cases:
  • Reasoning.
  • Function calling.
  • Subject matter experts via fine-tuning.
  • Multilingual instruction following.
  • Translation.
• Out-of-scope: Use in any manner that violates applicable laws or regulations (including trade compliance laws).
• Release Date: 04/29/2026
• Version: 1.0
• Model Developers: RedHat (Neural Magic)

Model Optimizations

This model was obtained by quantizing the weights of MiniMax-M2.5 to INT8 data type. This optimization reduces the number of bits used to represent weights and activations from 16 to 8, reducing GPU memory requirements (by approximately 50%) and increasing matrix-multiply compute throughput (by approximately 2x). Weight quantization also reduces disk size requirements by approximately 50%.

Only weights and activations of the linear operators within transformers blocks are quantized. Weights are quantized with a symmetric static per-channel scheme, whereas activations are quantized with a symmetric dynamic per-token scheme. A combination of the SmoothQuant and GPTQ algorithms is applied for quantization, as implemented in the llm-compressor library.

Deployment

This model can be deployed efficiently using the vLLM backend, as shown in the example below.

from vllm import LLM, SamplingParams
from transformers import AutoTokenizer

model_id = "RedHatAI/MiniMax-M2.5-quantized.w8a8"
number_gpus = 1
sampling_params = SamplingParams(temperature=1.0, top_p=0.95, top_k=40, min_p=0, max_tokens=256)

messages = [
    {"role": "user", "content": prompt}
]

tokenizer = AutoTokenizer.from_pretrained(model_id)

messages = [{"role": "user", "content": "Give me a short introduction to large language model."}]

prompts = tokenizer.apply_chat_template(messages, add_generation_prompt=True, tokenize=False)

llm = LLM(model=model_id, tensor_parallel_size=number_gpus)

outputs = llm.generate(prompts, sampling_params)

generated_text = outputs[0].outputs[0].text
print(generated_text)

vLLM aslo supports OpenAI-compatible serving. See the documentation for more details.

Creation

Creation details

This model was created with [llm-compressor](https://github.com/vllm-project/llm-compressor) by running the code snippet below.

from datasets import load_dataset
from transformers import AutoModelForCausalLM, AutoTokenizer, AutoProcessor
from llmcompressor import oneshot
from llmcompressor.modifiers.quantization import GPTQModifier

MODEL_ID = "RedHatAI/MiniMax-M2.5-BF16"

model = AutoModelForCausalLM.from_pretrained(MODEL_ID, torch_dtype="auto", trust_remote_code=True)
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID, trust_remote_code=True)
processor = AutoProcessor.from_pretrained(MODEL_ID)

NUM_CALIBRATION_SAMPLES=512
MAX_SEQUENCE_LENGTH=2048

# Load dataset.
ds = load_dataset("HuggingFaceH4/ultrachat_200k", split=f"train_sft[:{NUM_CALIBRATION_SAMPLES}]")
ds = ds.shuffle(seed=42)

# Preprocess the data into the format the model is trained with.
def preprocess(example):
    return {"text": tokenizer.apply_chat_template(example["messages"], tokenize=False)}

ds = ds.map(preprocess)

# Tokenize the data (be careful with bos tokens - we need add_special_tokens=False since the chat_template already added it).
def tokenize(sample):
    return tokenizer(sample["text"], padding=False, max_length=MAX_SEQUENCE_LENGTH, truncation=True, add_special_tokens=False)
ds = ds.map(tokenize, remove_columns=ds.column_names)

# Configure the quantization algorithm to run.
recipe = GPTQModifier( scheme="W8A8", weight_observer="mse", targets= [r"re:.*block_sparse_moe\.experts\.\d+\.w[1-3]$", r"re:.*mlp\.experts\.\d+\.(gate|up|gate_up|down)_proj$" ], ignore=["re:.*self_attn.*", "lm_head"])


# Apply quantization.
oneshot(
    model=model, dataset=ds,
    recipe=recipe,
    max_seq_length=MAX_SEQUENCE_LENGTH,
    num_calibration_samples=NUM_CALIBRATION_SAMPLES,
    processor=processor
)

# Save to disk compressed.
SAVE_DIR = MODEL_ID.rstrip("/").split("/")[-1] + ".w8a8"
model.save_pretrained(SAVE_DIR, save_compressed=True)
tokenizer.save_pretrained(SAVE_DIR)

Evaluation

The model was evaluated on the ifeval, mmlu_pro and gsm8k_platinum using lm-evaluation-harness, on reasoning tasks using lighteval. vLLM was used for all evaluations.

Evaluation details

Deploy using vllm to create an OpenAI-compatible API endpoint:

• vLLM:

  vllm serve RedHatAI/MiniMax-M2.5.w8a8 --max-model-len 262144 --reasoning-parser deepseek_r1
  

  lm-evaluation-harness

  lm_eval --model local-chat-completions \
    --tasks mmlu_pro_chat \
    --model_args "model=RedHatAI/MiniMax-M2.5.w8a8,max_length=262144,base_url=http://0.0.0.0:8000/v1/chat/completions,num_concurrent=64,max_retries=3,tokenized_requests=False,tokenizer_backend=None,timeout=1200" \
    --num_fewshot 0 \
    --apply_chat_template \
    --gen_kwargs "do_sample=True,temperature=1.0,top_p=0.95,top_k=40,min_p=0.0,max_gen_toks=64000
  

  lm_eval --model local-chat-completions \
    --tasks ifeval \
    --model_args "model=RedHatAI/MiniMax-M2.5.w8a8,max_length=262144,base_url=http://0.0.0.0:8000/v1/chat/completions,num_concurrent=64,max_retries=3,tokenized_requests=False,tokenizer_backend=None,timeout=1200" \
    --num_fewshot 0 \
    --apply_chat_template \
    --gen_kwargs "do_sample=True,temperature=1.0,top_p=0.95,top_k=40,min_p=0.0,max_gen_toks=64000
  

  lm_eval --model local-chat-completions \
    --tasks gsm8k_platinum_cot_llama \
    --model_args "model=RedHatAI/MiniMax-M2.5.w8a8,max_length=262144,base_url=http://0.0.0.0:8000/v1/chat/completions,num_concurrent=64,max_retries=3,tokenized_requests=False,tokenizer_backend=None,timeout=1200" \
    --num_fewshot 0 \
    --apply_chat_template \
    --gen_kwargs "do_sample=True,temperature=1.0,top_p=0.95,top_k=40,min_p=0.0,max_gen_toks=64000
  

  lighteval

  lighteval_model_arguments.yaml

  model_parameters:
    model_name: RedHatAI/MiniMax-M2.5.w8a8
    dtype: auto
    gpu_memory_utilization: 0.9
    max_model_length: 40960
    generation_parameters:
      temperature: 1.0
      top_k: 40
      min_p: 0.0
      top_p: 0.95
      max_new_tokens: 64000
  

  lighteval endpoint litellm lighteval_model_arguments.yaml  \
    "aime25|0,math_500|0,gpqa:diamond|0"
  

Accuracy

Benchmark	RedHatAI/MiniMax-M2.5-BF16	RedHatAI/MiniMax-M2.5.w8a8	Recovery (%)
GSM8k Platinum (0-shot)	95.15	95.18	100.03
IfEval (0-shot)	92.05	90.33	98.13
AIME 2025	87.50	88.33	100.95
GPQA diamond	83.67	84.51	101.01
Math 500	87.33	87.13	99.77
MMLU Pro Chat	80.83	81.25	100.51