Skip to content
Figure 1: Antares 1B · diagram by Sebastian Raschka · click to zoom
DenseCisco 🇺🇸 · July 21, 2026

Antares (1B)

Cisco Foundation AI's compact terminal agent uses a dense Granite 4.0 backbone and is post-trained with SFT and GRPO for vulnerability localization.

Key detail

Retains Granite 4.0's MuP scaling multipliers and tied input/output embeddings, then adds SFT and GRPO for terminal-based vulnerability localization.

Parameters

1B

Context

128K

KV / token

80 KiB

Fact sheet

Scale
1B parameters
Context window
131,072 tokens
Decoder type
Dense
Attention
GQA with RoPE
Layer mix
40 GQA
KV cache / token (bf16)
80 KiB · Moderate
License
Apache License 2.0
Released

Efficiency

Parameters active per token

100%

Dense: every weight runs for every token.

KV cache per token (bf16)

80 KiB

Smaller than 52% of models in the atlas. How it’s calculated

Architecture concepts

Use the ← and → keys to step through releases.

Similar architectures

  • Architecture diagram of Llama 3.2 (1B)

    Meta, released 2024-09-25. Dense decoder, 1B parameters, 128K context. Attention: GQA.

  • Architecture diagram of Qwen3 (0.6B)

    Alibaba Cloud, released 2025-04-28. Dense decoder, 600M parameters, 32K context. Attention: GQA.

  • Architecture diagram of SmolLM3 (3B)

    Hugging Face, released 2025-06-19. Dense decoder, 3B parameters, 128K context. Attention: GQA with periodic NoPE layers.

  • Architecture diagram of Llama 3.2 (3B)

    Meta, released 2024-09-25. Dense decoder, 3B parameters, 128K context. Attention: GQA.

Search the atlas

Jump to a model, concept or page