Foundation AI

Antares 1B

Name: Antares 1B
Rating: 62 (196 reviews)
Author: Foundation AI

当前

HuggingFace

5.7K下载量151点赞Jul 2026发布日期131K tokens上下文Apache 2.0许可证42 基础质量

Antares 1B (1.840000033378601B parameters) requires approximately 4.1 GB of VRAM with Q4_K_M quantization. For the best balance of quality and speed, we recommend hardware with at least 5 GB of VRAM.

快速开始

— 复制粘贴即可本地运行

Copy-paste commands to run Antares 1B on your machine.

Run

docker run --rm -it ghcr.io/ggerganov/llama.cpp:full \
  --hf-repo "fdtn-ai/antares-1b" \
  --hf-file "antares-1b-Q4_K_M.gguf" \
  -c 4096 -ngl 99

Quick specs

Parameters1.8B

Architecturedense

Context131K tokens

Modalitytext

Min RAM0.7 GB

Rec. RAM1.1 GB (Q4_K_M)

LicenseApache 2.0

FamilyAntares

✓ Code

About this model

Antares 1B is Foundation AI's compact security-focused agentic model, reinforcement-learning post-trained on top of IBM's Granite 4.0 1B. It targets terminal-agent workflows and vulnerability detection at a size small enough for local and edge deployment, with a 131K context.

•1.84B parameters — runs locally on CPU or a small GPU.
•Specialised for security work: vulnerability detection and terminal-agent tasks.
•Reinforcement-learning post-trained on IBM Granite 4.0 1B.
•131,072-token context.
•Apache 2.0 licensed.

Antares 1B 的最佳选择

运行此模型

Antares 1B on RTX 2060 6GB Antares 1B on RTX 4050 Laptop 6GB Antares 1B on GTX 1060 6GB

量化选项

各量化级别的 VRAM 估算

No hardware detected — fit column shows raw VRAM estimates

Quant	Bits	VRAM	Quality	Fit
Q1_0_G128	1.125	0.3 GB	Very Low	—
Q2_0_G128	1.71	0.5 GB	Low	—
Q2_K	2	0.7 GB	Low	—
Q3_K_S	3	0.9 GB	Low	—
NVFP4	4	1.0 GB	Medium	—
Q4_K_M	4	1.1 GB	Medium	—
Q5_K_M	5	1.3 GB	High	—
Q6_K	6	1.5 GB	High	—
Q8_0	8	2.0 GB	Very High	—
F16	16	3.8 GB	Maximum	—

硬件兼容性

全部硬件的适配估算

打开计算器

Computing compatibility...

内存详细分析

Reference: RTX 2060 6GB

Weights1.1 GB

KV Cache1.2 GB

Runtime1.2 GB

Headroom0.6 GB

常见问题

FAQ — Antares 1B

另请参阅

量化指南评分方法打开计算器