Instructions to use mlx-community/LongCat-Flash-Thinking-ZigZag-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use mlx-community/LongCat-Flash-Thinking-ZigZag-4bit with MLX:

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

# Generate text with mlx-lm
from mlx_lm import load, generate

model, tokenizer = load("mlx-community/LongCat-Flash-Thinking-ZigZag-4bit")

prompt = "Write a story about Einstein"
messages = [{"role": "user", "content": prompt}]
prompt = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True
)

text = generate(model, tokenizer, prompt=prompt, verbose=True)

Transformers

How to use mlx-community/LongCat-Flash-Thinking-ZigZag-4bit with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="mlx-community/LongCat-Flash-Thinking-ZigZag-4bit", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, LongcatCausalLM

tokenizer = AutoTokenizer.from_pretrained("mlx-community/LongCat-Flash-Thinking-ZigZag-4bit", trust_remote_code=True)
model = LongcatCausalLM.from_pretrained("mlx-community/LongCat-Flash-Thinking-ZigZag-4bit", trust_remote_code=True, device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings
LM Studio

vLLM

How to use mlx-community/LongCat-Flash-Thinking-ZigZag-4bit with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "mlx-community/LongCat-Flash-Thinking-ZigZag-4bit"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mlx-community/LongCat-Flash-Thinking-ZigZag-4bit",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/mlx-community/LongCat-Flash-Thinking-ZigZag-4bit

SGLang

How to use mlx-community/LongCat-Flash-Thinking-ZigZag-4bit with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "mlx-community/LongCat-Flash-Thinking-ZigZag-4bit" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mlx-community/LongCat-Flash-Thinking-ZigZag-4bit",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "mlx-community/LongCat-Flash-Thinking-ZigZag-4bit" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mlx-community/LongCat-Flash-Thinking-ZigZag-4bit",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

How to use mlx-community/LongCat-Flash-Thinking-ZigZag-4bit with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "mlx-community/LongCat-Flash-Thinking-ZigZag-4bit"

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": "mlx-community/LongCat-Flash-Thinking-ZigZag-4bit"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

MLX LM

How to use mlx-community/LongCat-Flash-Thinking-ZigZag-4bit with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "mlx-community/LongCat-Flash-Thinking-ZigZag-4bit"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "mlx-community/LongCat-Flash-Thinking-ZigZag-4bit"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "mlx-community/LongCat-Flash-Thinking-ZigZag-4bit",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Docker Model Runner
How to use mlx-community/LongCat-Flash-Thinking-ZigZag-4bit with Docker Model Runner:
```
docker model run hf.co/mlx-community/LongCat-Flash-Thinking-ZigZag-4bit
```

Hermes Agent

How to use mlx-community/LongCat-Flash-Thinking-ZigZag-4bit 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 "mlx-community/LongCat-Flash-Thinking-ZigZag-4bit"

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 mlx-community/LongCat-Flash-Thinking-ZigZag-4bit

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use mlx-community/LongCat-Flash-Thinking-ZigZag-4bit with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "mlx-community/LongCat-Flash-Thinking-ZigZag-4bit"

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 "mlx-community/LongCat-Flash-Thinking-ZigZag-4bit" \
  --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"

LongCat-Flash-Thinking-ZigZag-4bit / chat_template.jinja

kernelpool

Update chat_template.jinja

51c9ce1 verified 9 months ago

Raw History Blame Contribute Delete

7.75 kB

{%- set tool_choice = tool_choice | default('auto') %}
{%- set enable_thinking = enable_thinking | default(true) %}
{%- set save_history_reasoning_content = save_history_reasoning_content | default(false) %}
{%- set ns = namespace(tool_types = [], last_query_index = -1) %}
{%- for idx in range(messages|length - 1) %}
    {%- set msg = messages[idx] %}
    {%- if msg.role == 'assistant' and not msg.tool_calls %}
        {%- set ns.last_query_index = idx %}
    {%- endif %}
{%- endfor %}
{%- if tools and tool_choice != 'none' %}
    {{- "<longcat_tool_declare>\n"-}}
    {{- "# Tools\n" }}
    {{- "You have access to the following tools:\n\n" }}
    {%- for tool in tools %}
        {%- if tool.type not in ns.tool_types %}
            {%- set ns.tool_types = ns.tool_types + [tool.type] %}
            {{- "## Tool namespace: " ~ tool.type ~ "\n\n" }}
        {%- endif %}
        {%- if tool.type == 'code_interpreter' %}
            {%- set tool = {"type":"code_interpreter","function":{"name":"code_interpreter_preview","description":"The code will be executed in a stateful Jupyter notebook sandbox environment, only supports local computation, data processing, and file operations.\nCode sandbox environment (network isolated) Any external network requests or online API calls are prohibited.\nIf online functionality is needed, please use other permitted tools.\nCode will respond with the output of the execution or time out after 60.0 seconds. ","parameters":{"type":"object","properties":{"language":{"type":"string","description":"The programming language of the code to be executed. Available values: python (Default), java, go, js, ts, c, c++."},"code":{"type":"string","description":"Python code to be executed must not include the following:\n- Importing network libraries such as requests, httplib, etc.\n- Any form of HTTP requests.\n- External API calls.\n- Network port operations. Example: ```python\nimport pandas as pd\npd.DataFrame({'A':[1,2]})\n```"},"timeout":{"type":"number","description":"The maximum execution time of the code, in seconds. Default is 60.0."}}},"required":["code"]}} %}
        {%- endif %}
        {{- "### Tool name: " + tool.function.name + "\n" }}
        {{- "Description: " + tool.function.description + "\n\n" }}
        {{- "InputSchema: " + tool.function.parameters | tojson(ensure_ascii=False) + "\n\n" }}
    {%- endfor %}
    {{- '**Note**: For each function call, output the function name and arguments within the following XML format:\n<longcat_tool_call>{function-name}\n<longcat_arg_key>{arg-key-1}</longcat_arg_key>\n<longcat_arg_value>{arg-value-1}</longcat_arg_value>\n<longcat_arg_key>{arg-key-2}</longcat_arg_key>\n<longcat_arg_value>{arg-value-2}</longcat_arg_value>\n...\n</longcat_tool_call>\n' }}
    {{- "</longcat_tool_declare>"-}}
{%- endif %}
{%- for msg in messages %}
    {%- if msg.role == "system" %}
        {{- "<longcat_system>" + msg.content }}
    {%- elif msg.role == "user" %}
        {{- "<longcat_user>" }}
        {%- if msg["files"] %}
            {{- '<longcat_files>\n' ~ msg.files | tojson(indent=2) ~ '\n</longcat_files>' }}
        {%- endif %}
        {{- msg.content }}
        {%- if save_history_reasoning_content and enable_thinking is not none %}
            {%- if loop.index0 == messages|length - 1 %}
                {# pass #}
            {%- else %}
                {%- if enable_thinking == true %}
                    {{- " /think_on" }}
                    {%- if thinking_budget %}
                        {%- if thinking_budget < 1024 %}
                            {%- set thinking_budget = 1024 %}
                        {%- endif %}
                        {{- "\nthinking_budget: < " ~ thinking_budget ~ "." }}
                    {%- endif %}
                    {{- " " }}
                {%- elif enable_thinking == false %}
                    {{- " /think_off " }}
                {%- endif %}
            {%- endif %}
        {%- elif loop.index0 > ns.last_query_index and enable_thinking is not none %}
            {%- if loop.index0 == messages|length - 1 %}
                {# pass #}
            {%- else %}
                {%- if enable_thinking == true %}
                    {{- " /think_on" }}
                    {%- if thinking_budget %}
                        {%- if thinking_budget < 1024 %}
                            {%- set thinking_budget = 1024 %}
                        {%- endif %}
                        {{- "\nthinking_budget: < " ~ thinking_budget ~ "." }}
                    {%- endif %}
                    {{- " " }}
                {%- elif enable_thinking == false %}
                    {{- " /think_off " }}
                {%- endif %}
            {%- endif %}
        {%- endif %}
    {%- elif msg.role == "assistant" %}
        {{- "<longcat_assistant>" }}
        {%- if save_history_reasoning_content == true or loop.index0 > ns.last_query_index %}
            {%- if enable_thinking == true %}
                {%- if msg.reasoning_content %}
                    {{- "<longcat_think>\n" ~ msg.reasoning_content ~ "\n</longcat_think>\n" }}
                {%- else %}
                    {{- "<longcat_think>\n\n</longcat_think>\n" }}
                {%- endif %}
            {%- elif enable_thinking == false %}
                {{- "<longcat_think>\n\n</longcat_think>\n" }}
            {%- endif %}
        {%- endif %}
        {%- if msg.content %}
            {{- msg.content }}
        {%- endif %}
        {%- if msg.tool_calls %}
            {%- for tool_call in msg.tool_calls -%}
                {{- "<longcat_tool_call>" ~ tool_call.function.name ~ "\n" -}}
                {% set _args = tool_call.function.arguments %}
                {% for k, v in _args.items() %}
                    {{- "<longcat_arg_key>" ~ k ~ "</longcat_arg_key>\n" -}}
                    {{- "<longcat_arg_value>" ~ (v if v is string else v | tojson(ensure_ascii=False)) ~ "</longcat_arg_value>\n" -}}
                {% endfor %}
                {{- "</longcat_tool_call>\n" }}
            {%- endfor %}
        {%- endif %}
        {{- "</longcat_s>" -}}
    {%- elif msg.role == "tool" %}
        {%- if messages[loop.index0 - 1].role != "tool" %}
            {{- "<longcat_observation>" -}}
        {%- endif %}
        {{- "<longcat_tool_response>" ~ msg.name ~ "\n" ~ msg.content ~ "</longcat_tool_response>"-}}
        {%- if loop.index0 == messages|length - 1 %}
            {# pass #}
        {%- elif save_history_reasoning_content or loop.index0 > ns.last_query_index %}
            {%- if enable_thinking == true and messages[loop.index0 + 1].role != "tool" %}
                {{- " /think_on" }}
                {%- if thinking_budget %}
                    {%- if thinking_budget < 1024 %}
                        {%- set thinking_budget = 1024 %}
                    {%- endif %}
                    {{- "\nthinking_budget: < " ~ thinking_budget ~ "." }}
                {%- endif %}
                {{- " " }}
            {%- elif enable_thinking == false %}
                {{- " /think_off " }}
            {%- endif %}
        {%- endif %}
    {%- endif %}
{%- endfor %}
{%- if add_generation_prompt %}
    {%- if enable_thinking == true %}
        {{- " /think_on" }}
        {%- if thinking_budget %}
            {%- if thinking_budget < 1024 %}
                {%- set thinking_budget = 1024 %}
            {%- endif %}
            {{- "\nthinking_budget: < " ~ thinking_budget ~ "." }}
        {%- endif %}
        {{- " <longcat_assistant><longcat_think>\n" }}
    {%- elif enable_thinking == false %}
        {{- " /think_off <longcat_assistant><longcat_think>\n\n</longcat_think>\n" }}
    {%- else %}
        {{- "<longcat_assistant>" }}
    {%- endif %}
{%- endif %}

	{%- set tool_choice = tool_choice \| default('auto') %}
	{%- set enable_thinking = enable_thinking \| default(true) %}
	{%- set save_history_reasoning_content = save_history_reasoning_content \| default(false) %}
	{%- set ns = namespace(tool_types = [], last_query_index = -1) %}
	{%- for idx in range(messages\|length - 1) %}
	{%- set msg = messages[idx] %}
	{%- if msg.role == 'assistant' and not msg.tool_calls %}
	{%- set ns.last_query_index = idx %}
	{%- endif %}
	{%- endfor %}

	{%- if tools and tool_choice != 'none' %}
	{{- "<longcat_tool_declare>\n"-}}
	{{- "# Tools\n" }}
	{{- "You have access to the following tools:\n\n" }}
	{%- for tool in tools %}
	{%- if tool.type not in ns.tool_types %}
	{%- set ns.tool_types = ns.tool_types + [tool.type] %}
	{{- "## Tool namespace: " ~ tool.type ~ "\n\n" }}
	{%- endif %}
	{%- if tool.type == 'code_interpreter' %}
	{%- set tool = {"type":"code_interpreter","function":{"name":"code_interpreter_preview","description":"The code will be executed in a stateful Jupyter notebook sandbox environment, only supports local computation, data processing, and file operations.\nCode sandbox environment (network isolated) Any external network requests or online API calls are prohibited.\nIf online functionality is needed, please use other permitted tools.\nCode will respond with the output of the execution or time out after 60.0 seconds. ","parameters":{"type":"object","properties":{"language":{"type":"string","description":"The programming language of the code to be executed. Available values: python (Default), java, go, js, ts, c, c++."},"code":{"type":"string","description":"Python code to be executed must not include the following:\n- Importing network libraries such as requests, httplib, etc.\n- Any form of HTTP requests.\n- External API calls.\n- Network port operations. Example: ```python\nimport pandas as pd\npd.DataFrame({'A':[1,2]})\n```"},"timeout":{"type":"number","description":"The maximum execution time of the code, in seconds. Default is 60.0."}}},"required":["code"]}} %}
	{%- endif %}
	{{- "### Tool name: " + tool.function.name + "\n" }}
	{{- "Description: " + tool.function.description + "\n\n" }}
	{{- "InputSchema: " + tool.function.parameters \| tojson(ensure_ascii=False) + "\n\n" }}
	{%- endfor %}
	{{- 'Note: For each function call, output the function name and arguments within the following XML format:\n<longcat_tool_call>{function-name}\n<longcat_arg_key>{arg-key-1}</longcat_arg_key>\n<longcat_arg_value>{arg-value-1}</longcat_arg_value>\n<longcat_arg_key>{arg-key-2}</longcat_arg_key>\n<longcat_arg_value>{arg-value-2}</longcat_arg_value>\n...\n</longcat_tool_call>\n' }}
	{{- "</longcat_tool_declare>"-}}
	{%- endif %}

	{%- for msg in messages %}
	{%- if msg.role == "system" %}
	{{- "<longcat_system>" + msg.content }}
	{%- elif msg.role == "user" %}
	{{- "<longcat_user>" }}
	{%- if msg["files"] %}
	{{- '<longcat_files>\n' ~ msg.files \| tojson(indent=2) ~ '\n</longcat_files>' }}
	{%- endif %}
	{{- msg.content }}
	{%- if save_history_reasoning_content and enable_thinking is not none %}
	{%- if loop.index0 == messages\|length - 1 %}
	{# pass #}
	{%- else %}
	{%- if enable_thinking == true %}
	{{- " /think_on" }}
	{%- if thinking_budget %}
	{%- if thinking_budget < 1024 %}
	{%- set thinking_budget = 1024 %}
	{%- endif %}
	{{- "\nthinking_budget: < " ~ thinking_budget ~ "." }}
	{%- endif %}
	{{- " " }}
	{%- elif enable_thinking == false %}
	{{- " /think_off " }}
	{%- endif %}
	{%- endif %}
	{%- elif loop.index0 > ns.last_query_index and enable_thinking is not none %}
	{%- if loop.index0 == messages\|length - 1 %}
	{# pass #}
	{%- else %}
	{%- if enable_thinking == true %}
	{{- " /think_on" }}
	{%- if thinking_budget %}
	{%- if thinking_budget < 1024 %}
	{%- set thinking_budget = 1024 %}
	{%- endif %}
	{{- "\nthinking_budget: < " ~ thinking_budget ~ "." }}
	{%- endif %}
	{{- " " }}
	{%- elif enable_thinking == false %}
	{{- " /think_off " }}
	{%- endif %}
	{%- endif %}
	{%- endif %}
	{%- elif msg.role == "assistant" %}
	{{- "<longcat_assistant>" }}
	{%- if save_history_reasoning_content == true or loop.index0 > ns.last_query_index %}
	{%- if enable_thinking == true %}
	{%- if msg.reasoning_content %}
	{{- "<longcat_think>\n" ~ msg.reasoning_content ~ "\n</longcat_think>\n" }}
	{%- else %}
	{{- "<longcat_think>\n\n</longcat_think>\n" }}
	{%- endif %}
	{%- elif enable_thinking == false %}
	{{- "<longcat_think>\n\n</longcat_think>\n" }}
	{%- endif %}
	{%- endif %}
	{%- if msg.content %}
	{{- msg.content }}
	{%- endif %}
	{%- if msg.tool_calls %}
	{%- for tool_call in msg.tool_calls -%}
	{{- "<longcat_tool_call>" ~ tool_call.function.name ~ "\n" -}}
	{% set _args = tool_call.function.arguments %}
	{% for k, v in _args.items() %}
	{{- "<longcat_arg_key>" ~ k ~ "</longcat_arg_key>\n" -}}
	{{- "<longcat_arg_value>" ~ (v if v is string else v \| tojson(ensure_ascii=False)) ~ "</longcat_arg_value>\n" -}}
	{% endfor %}
	{{- "</longcat_tool_call>\n" }}
	{%- endfor %}
	{%- endif %}
	{{- "</longcat_s>" -}}
	{%- elif msg.role == "tool" %}
	{%- if messages[loop.index0 - 1].role != "tool" %}
	{{- "<longcat_observation>" -}}
	{%- endif %}
	{{- "<longcat_tool_response>" ~ msg.name ~ "\n" ~ msg.content ~ "</longcat_tool_response>"-}}
	{%- if loop.index0 == messages\|length - 1 %}
	{# pass #}
	{%- elif save_history_reasoning_content or loop.index0 > ns.last_query_index %}
	{%- if enable_thinking == true and messages[loop.index0 + 1].role != "tool" %}
	{{- " /think_on" }}
	{%- if thinking_budget %}
	{%- if thinking_budget < 1024 %}
	{%- set thinking_budget = 1024 %}
	{%- endif %}
	{{- "\nthinking_budget: < " ~ thinking_budget ~ "." }}
	{%- endif %}
	{{- " " }}
	{%- elif enable_thinking == false %}
	{{- " /think_off " }}
	{%- endif %}
	{%- endif %}
	{%- endif %}
	{%- endfor %}
	{%- if add_generation_prompt %}
	{%- if enable_thinking == true %}
	{{- " /think_on" }}
	{%- if thinking_budget %}
	{%- if thinking_budget < 1024 %}
	{%- set thinking_budget = 1024 %}
	{%- endif %}
	{{- "\nthinking_budget: < " ~ thinking_budget ~ "." }}
	{%- endif %}
	{{- " <longcat_assistant><longcat_think>\n" }}
	{%- elif enable_thinking == false %}
	{{- " /think_off <longcat_assistant><longcat_think>\n\n</longcat_think>\n" }}
	{%- else %}
	{{- "<longcat_assistant>" }}
	{%- endif %}
	{%- endif %}