DeepSeek-V4-Flash-0731-GGUF

All projects

Apps & Tools

DeepSeek-V4-Flash-0731-GGUF

High-performance GGUF quants of DeepSeek-V4-Flash with enhanced agentic capabilities

byUnsloth AI

Playable

DeepSeek-V4-Flash-0731-GGUF media is blocked

Allow external media to connect to the provider and play this content.

Description

Quantized versions of the DeepSeek-V4-Flash model using Unsloth's Dynamic 2.0 technique for superior accuracy in local inference. This release features enhanced agentic performance and speculative decoding via the DSpark module, which allows for up to 2x faster decoding while supporting a wide range of bits from 1-bit to 16-bit.

Descriptions, tags, and model credits may be AI-generated or inferred from public sources and can be incomplete or wrong. Learn more.

Details

Why this is interesting: A significant local-inference release featuring state-of-the-art quantization and speculative decoding for a top-tier…
Attributed to: @unsloth
Built with: Gemini 3.7 Flash DeepSeek-V4-Flash-0731 Qwen3.7 Flash 3.6 Flash 3.5 Flash Cyber Step 3.7 Flash Gemini 3.5 Flash V4 2.5 Flash
Links: huggingface.co
View tweet
Tags: #quantization #gguf #llm #agent

Discovered: X
First posted: X · Aug 2, 2026
Listed: Added 45d ago
Build: Strong
Models: Strong