What are the hardware requirements for Agents K1?

Agents K1 requires approximately 8.8 GB of VRAM and supports a context window of 256K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Agents K1 and how large is it?

Agents K1 is developed by InternScience, a model with 4b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate Agents K1?

Agents K1 is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Agents K1 by InternScience — VRAM 8.8GB, 256K context

Name: Agents K1
Author: InternScience

Agents K1 is an open-source language model by InternScience. Features: 4b LLM, VRAM: 8.8GB, Context: 256K, License: apache-2.0, Instruction-Based, LLM Explorer Score: 0.28.

Base model:finetune:qwen/qwen3... Base model:qwen/qwen3-4b-instr... Biomedical Conversational En Endpoints compatible Grpo Information-extraction Instruct Named-entity-recognition Qwen3 Region:us Relation-extraction Safetensors Scientific-text Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/InternScience/Agents-K1

Agents K1 Parameters and Internals

LLM Name	Agents K1
Repository 🤗	https://huggingface.co/InternScience/Agents-K1
Base Model(s)	Qwen3 4B Instruct 2507 Qwen/Qwen3-4B-Instruct-2507
Model Size	4b
Required VRAM	8.8 GB
Updated	2026-07-16
Maintainer	InternScience
Model Type	qwen3
Instruction-Based	Yes
Model Files	5.0 GB: 1-of-2 3.8 GB: 2-of-2
Supported Languages	en
Model Architecture	Qwen3ForCausalLM
License	apache-2.0
Context Length	262144
Model Max Length	262144
Transformers Version	4.52.4
Tokenizer Class	Qwen2Tokenizer
Padding Token	<\|endoftext\|>
Vocabulary Size	151936
Torch Data Type	bfloat16
Errors	replace

Rank the Agents K1 capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to Agents K1

Best Alternatives	Context / RAM	Downloads	Likes
Fable Traces	256K / 8.1 GB	5530	209
FastContext 1.0 4B SFT	256K / 8.1 GB	5735	357
Qwen3 4B Instruct 2507	256K / 8.1 GB	3109972	911
FastContext 1.0 4B RL	256K / 8.1 GB	4559	61
Qwen3 4B Instruct 2507 FP8	256K / 5.2 GB	1100744	78
Neuron 4B Instruct	256K / 8.1 GB	313	1
Tac 1	256K / 8.1 GB	774	0
Nexa AI 4B Instruct	256K / 8 GB	576	2
FastContext 1.0 4B SFT	256K / 8.1 GB	273	18
Karnak 6B V1.0	256K / 11.9 GB	1098	0

Note: green Score (e.g. "73.2") means that the model is better than InternScience/Agents-K1.