LocalClaw/LLM/Agents-A1 4B
Open-weight local LLM

Agents-A1 4B

InternScience compact dense agent model with Apache 2.0 licensing, 262K context and official Q4_K_M GGUF artifacts for 8GB-class local assistants.

Laptop ready 8 GB RAM Q4_K_M Small local research assistant
Run with LocalClaw Compare all models
Parameters
4B
Minimum RAM
8 GB
Model size
2.71 GB
Quantization
Q4_K_M

Can Agents-A1 4B run locally?

Agents-A1 4B is a good fit for normal laptops and compact desktops with 8 GB RAM or more.

Search for agents-a1-4b in LM Studio or another GGUF-compatible runtime.

Model sourceInternScience/Agents-A1-4B-Q4_K_M-GGUF
chatcodereasoningvisionagentictool-callinglightlong-context

Install path

01
Check RAM fitMinimum 8 GB RAM. Start with the Q4_K_M quant.
02
Load the modelSearch agents-a1-4b in LM Studio.
03
Control locallyUse LocalClaw to manage models, agents, chat, channels and scheduled OpenClaw work.

Strengths

  • Official InternScience Apache 2.0 base release
  • Official Q4_K_M GGUF release is about 2.71GB
  • Runs on a single GPU in the official vLLM and SGLang examples
  • 262K context window for long local assistant sessions
  • Strong vendor-reported agent, instruction-following and research benchmark results for a 4B model
  • Official local-app card includes llama.cpp, LM Studio, Jan, Docker and llama-cpp-python paths

Limitations

  • The model is optimized for agentic workflows rather than pure chat polish
  • Vision and very long context can raise practical memory needs above the GGUF file size
  • Most performance claims are currently from the model authors
  • Runtime support for the newest Qwen3.5-style multimodal stack may require recent local apps

Best use cases

  • Small local research assistant
  • 8GB-class coding helper
  • Tool-use agent experiments
  • Long-context instruction following
  • Local multimodal assistant testing
  • Private lightweight reasoning workflows

Capability profile

speed
9
quality
7
coding
8
reasoning
8

Technical notes

Developer
InternScience
License
Apache 2.0
Context window
262,144 tokens
Architecture
Dense 4B agentic vision-language model based on the Agents-A1 family, with Qwen3.5-style architecture, 262K context and tool-use oriented chat formatting.

This model fits these next steps

Hardware fit is based on LocalClaw's RAM tier, model size and quantization metadata. Always leave memory headroom for your OS and runtime.

Entry laptop fitMacBook Air 8GBMore headroomMac mini M4 16GBAll compatible picks8GB RAM guide

Similar models to compare

Agents-A1 35B (3B active, MoE)MiniCPM5 1B 1BQwen 3 (4B) 4BLFM2.5-8B-A1B 8.3B (1.5B active)

Where to go next