Mistral 7B Instruct (Q4)

generalApache 2.0tools

Mistral AI's retired 7B Instruct v0.3 model, retained for compatibility with existing installs. Prefer a current small model for new work.

At a glance

Parameters: 7B
Quantization: Q4_K_M
Context window: 33K tokens
Approx. size: 4.4 GB
Engines: llama.cpp, MLX, Ollama
License: Apache-2.0 (open)
Version: v1.1.1 , released 2026-08-15
Family: mistral

Tuning

Defaults a gezel client applies out of the box, from real evaluation runs against this exact quantization.

sampling.temperature: 0.7
sampling.topP: 0.95
sampling.maxTokens: 8192
sampling.repetitionPenalty: 1.1
sampling.repetitionContext: 20
profiles: thinking-general, thinking-coding, thinking-precise, instruct, creative

Model behaviors

Client-side behavior modules the catalog enables for this model (reasoning-tag handling, fabrication detection, prompt shaping).

fabrication.detect-past-tense-no-tools
fabrication.detect-claim-without-tool
prompt.tool-cookbook-condensed
turn.preamble-folding
turn.ramble-detection
provider.flatten-tool-transcript
mcp.compact-tool-schemas
mcp.relax-required-fields
mcp.default-missing-fields

Sources

llama.cpp (GGUF): bartowski/Mistral-7B-Instruct-v0.3-GGUF · Mistral-7B-Instruct-v0.3-Q4_K_M.gguf · Q4_K_M
MLX: mlx-community/Mistral-7B-Instruct-v0.3-4bit
Ollama: mistral:7b
Upstream: https://ollama.com/library/mistral
Manifest: View on GitHub

Tags

mistral
tools
retired