You can deploy this model and start earning money today!

DuoNeural/
Mistral-7B-Instruct-v0.3-GGUF

Name: Mistral-7B-Instruct-v0.3-GGUF
Availability: InStock

mistralai/Mistral-7B-Instruct-v0.3

Views

No views yet

Mistral-7B-Instruct-v0.3 — GGUF Quants

Quantized GGUF versions of mistralai/Mistral-7B-Instruct-v0.3 — Mistral AI's flagship 7B instruction-tuned model. v0.3 adds function calling support and an improved tokenizer (v3 with 32768 vocabulary expansion) over v0.2. Mistral-7B remains one of the most capable and widely deployed 7B models for general-purpose inference.

Available Files

File	Quant	Size	Use Case
`Mistral-7B-Instruct-v0.3-Q8_0.gguf`	Q8_0	~7.2GB	Maximum quality
`Mistral-7B-Instruct-v0.3-Q6_K.gguf`	Q6_K	~5.5GB	Near-lossless
`Mistral-7B-Instruct-v0.3-Q5_K_M.gguf`	Q5_K_M	~4.8GB	High quality
`Mistral-7B-Instruct-v0.3-Q4_K_M.gguf`	Q4_K_M	~4.1GB	Recommended default
`Mistral-7B-Instruct-v0.3-Q3_K_M.gguf`	Q3_K_M	~3.3GB	Low VRAM
`Mistral-7B-Instruct-v0.3-IQ4_XS.gguf`	IQ4_XS	~3.6GB	Imatrix 4-bit
`Mistral-7B-Instruct-v0.3-IQ3_XXS.gguf`	IQ3_XXS	~2.7GB	Imatrix 3-bit
`Mistral-7B-Instruct-v0.3-IQ2_M.gguf`	IQ2_M	~2.4GB	Imatrix 2-bit
`Mistral-7B-Instruct-v0.3-IQ1_S.gguf`	IQ1_S	~1.6GB	Extreme compression
`Mistral-7B-Instruct-v0.3-fp16.gguf`	FP16	~14.0GB	Full precision
`imatrix.dat`	—	—	Importance matrix

Usage

bash

# llama.cpp (v0.3 uses [INST] format)
./llama-cli -m Mistral-7B-Instruct-v0.3-Q4_K_M.gguf \
  --ctx-size 8192 -n 512 \
  -p "[INST] Hello! [/INST]"

# Ollama
ollama run hf.co/DuoNeural/Mistral-7B-Instruct-v0.3-GGUF:Q4_K_M

About Mistral-7B-Instruct-v0.3

Parameters: 7B
Context: 32K tokens (with sliding window attention)
Architecture: Mistral (GQA, SWA, RoPE)
License: Apache 2.0
New in v0.3: Function calling, tokenizer v3 (32768 vocab), improved tool use

One of the most production-proven 7B models in the open-source ecosystem. Excellent choice for general-purpose inference, function calling pipelines, and as a fine-tuning base.

Quantized by DuoNeural using llama.cpp on RTX 5090.

DuoNeural

DuoNeural is an open AI research lab — human + AI in collaboration.

Platform	Link
HuggingFace	huggingface.co/DuoNeural
Website	duoneural.com
GitHub	github.com/DuoNeural
X / Twitter	@DuoNeural
Email	duoneural@proton.me
Newsletter	duoneural.beehiiv.com
Support	buymeacoffee.com/duoneural

DuoNeural Research Publications

Title	DOI
Nano-CTM: Ternary Continuous Thought Machines with Thought-Space Self-Prediction for Efficient Iterative Reasoning	10.5281/zenodo.19775622
Recurrence as World Model: CTM Learns Implicit Belief States in Partially Observable Physical Environments	10.5281/zenodo.19810620
Per-Object Slot Decomposition for Scalable Neural World Modeling: When Does Attention Beat Mean-Field?	10.5281/zenodo.19846804
The Dynamical Horizon Principle: CTM Gates Converge to the Predictability Limit of Dynamical Systems	10.5281/zenodo.19952612

Open access, CC BY 4.0. Authored by Archon, Jesse Caldwell, Aura — DuoNeural.