BF16 GGUF models

Unquantized bfloat16. Original quality and the largest size. 9004 models have a BF16 file. Sizes run from 1M to 2780B parameters. The longest trained context is 10240K tokens.

Most downloadedNewestLargestLongest contextSmallest
DE
Devstral Small 2505 Deepseek V3.2 Speciale Distill
DeepSeek · TeichAI
Params23.6B
Context128K
Q4_K_M14.3 GB
CodingReasoningTool use
QW
Qwen3.5 9B Distill Qwen 3.8 Flash 20261011
Qwen · kevinzhou21
Params9.2B
Context256K
Q4_K_M5.8 GB
CodingReasoningTool use
MI
Ministral 3 3B Reasoning 2512 Heretic
Mistral · coder3101
Params3.43B
Context256K
Q4_K_M2.1 GB
CodingReasoningTool use
SP
Spark X2.6 4b Abliterated
Spark · FrostWolfAI
Params4.11B
Context1024K
Q4_K_M2.6 GB
ReasoningTool use
QW
Qwen3-TTS-12Hz-1.7B-Base
Qwen · FILM6912
Params3.1B
Context—
Smallest1.7 GB
GGUF
YU
YuE2 RS 4.0x0.7
Yue · scragnog
Params3.63B
Context24K
Q4_K_M2.4 GB
GGUF
MI
MiniMax M3.1 Flash Q2 K
Minimax · smalinin
Params2.12B
Context1024K
Smallest3.9 GB
ReasoningTool useVision
QW
Qwen3.5-4B-Uncensored-HauhauCS-Aggressive-BF16
Qwen · gramparser
Params4.21B
Context256K
Q4_K_M2.7 GB
ReasoningTool useVision
IN
Internlm2_5 20b Chat
InternLM · legraphista
Params19.9B
Context32K
Smallest4.5 GB
GGUF
MI
MiroThinker v1.0 30B
Mirothinker · bartowski
Params30.5B
Context256K
Q4_K_M18.6 GB
ReasoningTool use
QW
Qwopus3.5 2B v3
Qwopus · Jackrong
Params1.88B
Context256K
Q4_K_M1.3 GB
ReasoningTool useVision
QI
Qianfan OCR
Qianfan · Reza2kn
Params4.42B
Context32K
Q4_K_M2.7 GB
ReasoningTool useVision
GR
Granite 4.0 350m Base
Granite · ibm-granite
Params352M
Context32K
Q4_K_M0.2 GB
GGUF
QW
Qwen3.6 35B A3B
Qwen · prithivMLmods
Params35.5B
Context256K
Q4_K_M21.7 GB
Vision
CL
clef
Clef · prithivMLmods
Params26.9B
Context256K
Q4_K_M16.5 GB
CodingReasoningTool use
FL
Flex.2 preview
Flex · Eviation
Params8.16B
Context—
Smallest1.7 GB
GGUF
RW
rwkv7-g1c-13.3b-20251231-ctx8192.pth
Rwkv · NaomiBTW
Params13.3B
Context1024K
Q4_K_M8.4 GB
GGUF
GE
humanizer-12b
Gemma · ericthewiz
Params11.9B
Context256K
Q4_K_M7.6 GB
Vision
GR
granite vision 3.3 2b
Granite · ibm-granite
Params2.53B
Context128K
Q4_K_M1.5 GB
Tool useVision
GL
GLM 5.3 Flash UNCENSORED
GLM · Solstice-AI
Params1.17B
Context1024K
Smallest1.3 GB
ReasoningTool useVision
QW
Qwen3 TTS 12Hz 0.6B CustomVoice
Qwen · BricksDisplay
Params652M
Context32K
Q4_K_M0.4 GB
GGUF
BI
bitnet embedding 0.6b
Bitnet · microsoft
Params596M
Context32K
Smallest0.4 GB
Embedding
QW
mango qwen27 orca bf16
Qwen · isilicon
Params27.3B
Context256K
Smallest54.7 GB
ReasoningTool use
VI
VibeThinker 1.5B
Vibethinker · prithivMLmods
Params1.78B
Context128K
Q4_K_M1.1 GB
Coding
QW
Qwen2.5-Omni-3B
Qwen · Mungert
Params3.4B
Context32K
Q4_K_M2.2 GB
GGUF
AG
AgentHorse 4B 1.0
Agents · CloudGoat
Params4.21B
Context256K
Q4_K_M2.7 GB
ReasoningTool useVision
QW
Qwen3.5 9B Heretic
Qwen · SassyDiffusion
Params8.95B
Context256K
Q4_K_M5.8 GB
ReasoningTool useVision
ST
StagAI CyberSecurity 26B A4B v2 VLM GGUF BF16
Stagai · deer-sec
Params25.2B
Context256K
Smallest50.5 GB
ReasoningTool useVision
ST
G4 31B Style r1 DPO c30
Style · ApocalypseParty
Params30.7B
Context256K
Q4_K_M18.7 GB
ReasoningTool use
QI
Qianfan-OCR
Qianfan · Mungert
Params4.42B
Context32K
Q4_K_M3.1 GB
ReasoningVision

Other quants

Keep browsing