# uses:
# - llama.cpp webui
# - llama.vim
# - pi
#
# target machines with 128GB+ VRAM:
# - Apple Mac Studio
# - Apple MacBook
# - NVIDIA DGX Spark

[*]
ctx-size             = 0
mmap                 = 1
kv-unified           = 1
parallel             = 4
spec-default         = 1

[qwen-test]
hf                   = unsloth/Qwen3.5-4B-GGUF:Q4_K_M
ctx-size             = 262144
batch-size           = 2048
ubatch-size          = 2048
top-p                = 1.0
top-k                = 0
min-p                = 0.01
temp                 = 1.0
chat-template-kwargs = {"reasoning_effort": "high"}

[gpt-oss-120b-hf]
hf                   = ggml-org/gpt-oss-120b-GGUF
ctx-size             = 262144
batch-size           = 2048
ubatch-size          = 2048
top-p                = 1.0
top-k                = 0
min-p                = 0.01
temp                 = 1.0
chat-template-kwargs = {"reasoning_effort": "high"}

[gpt-oss-120b-hf-low]
hf                   = ggml-org/gpt-oss-120b-GGUF
ctx-size             = 262144
batch-size           = 2048
ubatch-size          = 2048
top-p                = 1.0
top-k                = 0
min-p                = 0.01
temp                 = 1.0
chat-template-kwargs = {"reasoning_effort": "low"}

[gpt-oss-120b-hf-medium]
hf                   = ggml-org/gpt-oss-120b-GGUF
ctx-size             = 262144
batch-size           = 2048
ubatch-size          = 2048
top-p                = 1.0
top-k                = 0
min-p                = 0.01
temp                 = 1.0
chat-template-kwargs = {"reasoning_effort": "medium"}

[gpt-oss-120b-hf-high]
hf                   = ggml-org/gpt-oss-120b-GGUF
ctx-size             = 262144
batch-size           = 2048
ubatch-size          = 2048
top-p                = 1.0
top-k                = 0
min-p                = 0.01
temp                 = 1.0
chat-template-kwargs = {"reasoning_effort": "high"}

[qwen3-coder-30b-hf-fim]
hf                   = ggml-org/Qwen3-Coder-30B-A3B-Instruct-Q8_0-GGUF
ctx-size             = 65536
cache-reuse          = 256
batch-size           = 1024
ubatch-size          = 1024
alias                = fim,infill

[glm-4.7-flash-hf]
hf                   = ggml-org/GLM-4.7-Flash-GGUF:Q8_0
ctx-size             = 202752
batch-size           = 4096
ubatch-size          = 4096
top-p                = 1.0
top-k                = 0
temp                 = 0.2
chat-template-kwargs = {"enable_thinking": false}

[glm-4.7-flash-hf-thinking]
hf                   = ggml-org/GLM-4.7-Flash-GGUF:Q8_0
ctx-size             = 202752
batch-size           = 4096
ubatch-size          = 4096
top-p                = 0.95
top-k                = 0
temp                 = 1.0

[qwen3.6-35b-a3b-hf-think-spec]
hf                   = ggml-org/Qwen3.6-35B-A3B-GGUF:Q8_0
spec-draft-hf        = ggml-org/Qwen3.5-0.8B-GGUF:Q8_0
ctx-size             = 262144
batch-size           = 2048
ubatch-size          = 2048
alias                = qwen3.6-35b-a3b-hf-think-spec
chat-template-kwargs = {"preserve_thinking": true}

[qwen3.6-27b-hf-think-spec]
hf                   = ggml-org/Qwen3.6-27B-GGUF:Q8_0
spec-draft-hf        = ggml-org/Qwen3.5-0.8B-GGUF:Q8_0
ctx-size             = 262144
batch-size           = 2048
ubatch-size          = 2048
alias                = qwen3.6-27b-hf-think-spec
chat-template-kwargs = {"preserve_thinking": true}

# uses a dedicated, pure 4-bit drafter
[qwen3.6-35b-a3b-hf-think-mtp]
hf                   = ggml-org/Qwen3.6-35B-A3B-MTP-GGUF
spec-type            = draft-mtp
spec-draft-n-max     = 3
spec-draft-hf        = ggml-org/Qwen3.6-35B-A3B-GGUF
spec-draft-model     = mtp-Qwen3.6-35B-A3B-Q4_0.gguf
temp                 = 0.6
min-p                = 0.05
ctx-size             = 262144
batch-size           = 2048
ubatch-size          = 512
alias                = qwen3.6-35b-a3b-hf-think-mtp,inst
chat-template-kwargs = {"preserve_thinking": true}

# uses a dedicated, pure 4-bit drafter
[qwen3.6-27b-hf-think-mtp]
hf                   = ggml-org/Qwen3.6-27B-MTP-GGUF
spec-type            = draft-mtp
spec-draft-n-max     = 3
spec-draft-hf        = ggml-org/Qwen3.6-27B-GGUF
spec-draft-model     = mtp-Qwen3.6-27B-Q4_0.gguf
temp                 = 0.6
min-p                = 0.05
ctx-size             = 262144
batch-size           = 2048
ubatch-size          = 2048
alias                = qwen3.6-27b-hf-think-mtp,pi
reasoning-budget     = 4096
reasoning-budget-message = ... I am thinking for too long and cannot make a decision. I will now explain the problem to the user and ask them for advice.
chat-template-kwargs = {"preserve_thinking": true}

[gemma-4-26b-a4b-hf]
hf                   = ggml-org/gemma-4-26B-A4B-it-GGUF:Q8_0
ctx-size             = 262144
batch-size           = 2048
ubatch-size          = 2048
top-p                = 0.95
top-k                = 64
temp                 = 1.0
chat-template-kwargs = {"enable_thinking": false}

[gemma-4-31b-hf]
hf                   = ggml-org/gemma-4-31B-it-GGUF:Q8_0
ctx-size             = 262144
batch-size           = 2048
ubatch-size          = 2048
top-p                = 0.95
top-k                = 64
temp                 = 1.0
chat-template-kwargs = {"enable_thinking": false}

[gemma-4-26b-a4b-hf-think]
hf                   = ggml-org/gemma-4-26B-A4B-it-GGUF:Q8_0
ctx-size             = 262144
batch-size           = 2048
ubatch-size          = 2048
top-p                = 0.95
top-k                = 64
temp                 = 1.0
chat-template-kwargs = {"preserve_thinking": true}

[gemma-4-31b-hf-think]
hf                   = ggml-org/gemma-4-31B-it-GGUF:Q8_0
ctx-size             = 262144
batch-size           = 2048
ubatch-size          = 2048
top-p                = 0.95
top-k                = 64
temp                 = 1.0
chat-template-kwargs = {"preserve_thinking": true}