aquif-3.6-8B-Q4_K_M-GGUF

Edge-Quant/aquif-3.6-8B-Q4_K_M-GGUF

This model was converted to GGUF format from aquif-ai/aquif-3.6-8B using llama.cpp via the ggml.ai's GGUF-my-repo space. Refer to the original model card for more details on the model.

aquif-3.6-8B

Summary

aquif-3.6-8B is a hybrid reasoning model that automatically determines when and how deeply to think based on query complexity. Built on aquif-3.5-8B-Think with AutoThink RL data, it achieves 28% better token efficiency and 4% performance improvement across benchmarks.

Contents

• Key Features - Dynamic reasoning, efficiency gains, and smart resource allocation
• Performance - Benchmark results showing 4% average improvement
• Token Efficiency - 28% reduction in token usage
• Thinking Ratio - 12% reduction in thinking frequency
• Benchmark Highlights - Detailed results for AIME, LiveCodeBench, and GPQA Diamond
• Model Details - Architecture and specifications
• Usage - Code examples for implementation
• Previous Versions - Links to earlier models

Automatic Thinking

aquif-3.6-8B is a hybrid reasoning model that dynamically decides if and how much to think based on query complexity. Inspired by KAT-V1's approach of automatic thinking using AutoThink RL data on top of aquif-3.5-8B-Think, the model uses the following format:

<judge>
[analyzes whether to think or not]
</judge>

<think_on/off>
<think>
[thinking content]
</think>

<answer>
</answer>

This is the same format as KAT-V1-40B. Unlike something like DeepSeek-V3.1's toggleable reasoning that requires manual control (thinking_on/off), aquif-3.6's judge autonomously allocates reasoning depth - intelligently adapting its cognitive effort to each task automatically.

Key Features

• 🧠 Dynamic Reasoning: Automatically determines when and how deeply to think
• ⚡ 28% More Efficient: Significant token reduction while improving performance
• 📈 Better Performance: 4% average improvement across benchmarks
• 🎯 Smart Resource Allocation: 12% reduction in thinking ratio on average

Performance

<img src="https://cdn-uploads.huggingface.co/production/uploads/6747320df82ae35f0327cdd3/Jxv_kthSTqEiKdygHyVNl.png" alt="Chart 1" width="512"/>

Token Efficiency

<img src="https://cdn-uploads.huggingface.co/production/uploads/6747320df82ae35f0327cdd3/4k3HCx-LUigoGP51q_sdB.png" alt="Chart 2" width="512"/>

Thinking Ratio

<img src="https://cdn-uploads.huggingface.co/production/uploads/6747320df82ae35f0327cdd3/EkrkynHLQasZ3CRxySYsp.png" alt="Chart 3" width="512"/>

Benchmark Highlights

• AIME 2025: 26% fewer tokens, +1% performance, -7% thinking ratio
• LiveCodeBench: 32% fewer tokens, +4% performance, -18% thinking ratio
• GPQA Diamond: 24% fewer tokens, +6% performance, -11% thinking ratio

Model Details

• Base Model: 8B parameters
• Architecture: Hybrid reasoning with dynamic thinking allocation
• Context Length: 40K tokens
• License: Apache 2.0

Use with llama.cpp

Install llama.cpp through brew (works on Mac and Linux)

brew install llama.cpp

Invoke the llama.cpp server or the CLI.

CLI:

llama-cli --hf-repo Edge-Quant/aquif-3.6-8B-Q4_K_M-GGUF --hf-file aquif-3.6-8b-q4_k_m.gguf -p "The meaning to life and the universe is"

Server:

llama-server --hf-repo Edge-Quant/aquif-3.6-8B-Q4_K_M-GGUF --hf-file aquif-3.6-8b-q4_k_m.gguf -c 2048

Note: You can also use this checkpoint directly through the usage steps listed in the Llama.cpp repo as well.

Step 1: Clone llama.cpp from GitHub.

git clone https://github.com/ggerganov/llama.cpp

Step 2: Move into the llama.cpp folder and build it with LLAMA_CURL=1 flag along with other hardware-specific flags (for ex: LLAMA_CUDA=1 for Nvidia GPUs on Linux).

cd llama.cpp && LLAMA_CURL=1 make

Step 3: Run inference through the main binary.

./llama-cli --hf-repo Edge-Quant/aquif-3.6-8B-Q4_K_M-GGUF --hf-file aquif-3.6-8b-q4_k_m.gguf -p "The meaning to life and the universe is"

or

./llama-server --hf-repo Edge-Quant/aquif-3.6-8B-Q4_K_M-GGUF --hf-file aquif-3.6-8b-q4_k_m.gguf -c 2048