Scroll to navigation

LLAMA-SERVER(1) User Commands LLAMA-SERVER(1)

NAME

llama-server - llama-server

DESCRIPTION

----- common params -----

-h, --help, --usage print usage and exit

show version and build info

-cl, --cache-list show list of models in cache

print source-able bash completion script for llama.cpp

-t, --threads N number of CPU threads to use during generation (default: <num_cpus>)

(env: LLAMA_ARG_THREADS)

-tb, --threads-batch N number of threads to use during batch and prompt processing (default:

same as --threads)

-C, --cpu-mask M CPU affinity mask: arbitrarily long hex. Complements cpu-range

(default: "")

-Cr, --cpu-range lo-hi range of CPUs for affinity. Complements --cpu-mask

use strict CPU placement (default: 0)
set process/thread priority : low(-1), normal(0), medium(1), high(2), realtime(3) (default: 0)
use polling level to wait for work (0 - no polling, default: 50)

-Cb, --cpu-mask-batch M CPU affinity mask: arbitrarily long hex. Complements cpu-range-batch

(default: same as --cpu-mask)

-Crb, --cpu-range-batch lo-hi ranges of CPUs for affinity. Complements --cpu-mask-batch

use strict CPU placement (default: same as --cpu-strict)
set process/thread priority : 0-normal, 1-medium, 2-high, 3-realtime (default: 0)
use polling to wait for work (default: same as --poll)

-c, --ctx-size N size of the prompt context (default: 0, 0 = loaded from model)

(env: LLAMA_ARG_CTX_SIZE)

-n, --predict, --n-predict N number of tokens to predict (default: -1, -1 = infinity)

(env: LLAMA_ARG_N_PREDICT)

-b, --batch-size N logical maximum batch size (default: 2048)

(env: LLAMA_ARG_BATCH)

-ub, --ubatch-size N physical maximum batch size (default: 512)

(env: LLAMA_ARG_UBATCH)
number of tokens to keep from the initial prompt (default: 0, -1 = all)
use full-size SWA cache (default: false) [(more info)](https://github.com/ggml-org/llama.cpp/pull/13194#issuecomment-2868343055) (env: LLAMA_ARG_SWA_FULL)

-fa, --flash-attn [on|off|auto] set Flash Attention use ('on', 'off', or 'auto', default: 'auto')

(env: LLAMA_ARG_FLASH_ATTN)
whether to enable internal libllama performance timings (default: false) (env: LLAMA_ARG_PERF)

-e, --escape, --no-escape whether to process escapes sequences (\n, \r, \t, \', \", \\)

(default: true)
RoPE frequency scaling method, defaults to linear unless specified by the model (env: LLAMA_ARG_ROPE_SCALING_TYPE)
RoPE context scaling factor, expands context by a factor of N (env: LLAMA_ARG_ROPE_SCALE)
RoPE base frequency, used by NTK-aware scaling (default: loaded from model) (env: LLAMA_ARG_ROPE_FREQ_BASE)
RoPE frequency scaling factor, expands context by a factor of 1/N (env: LLAMA_ARG_ROPE_FREQ_SCALE)
YaRN: original context size of model (default: 0 = model training context size) (env: LLAMA_ARG_YARN_ORIG_CTX)
YaRN: extrapolation mix factor (default: -1.00, 0.0 = full interpolation) (env: LLAMA_ARG_YARN_EXT_FACTOR)
YaRN: scale sqrt(t) or attention magnitude (default: -1.00) (env: LLAMA_ARG_YARN_ATTN_FACTOR)
YaRN: high correction dim or alpha (default: -1.00) (env: LLAMA_ARG_YARN_BETA_SLOW)
YaRN: low correction dim or beta (default: -1.00) (env: LLAMA_ARG_YARN_BETA_FAST)

-kvo, --kv-offload, -nkvo, --no-kv-offload

(env: LLAMA_ARG_KV_OFFLOAD)
whether to enable weight repacking (default: enabled) (env: LLAMA_ARG_REPACK)
bypass host buffer allowing extra buffers to be used (env: LLAMA_ARG_NO_HOST)

-ctk, --cache-type-k TYPE KV cache data type for K

(default: f16) (env: LLAMA_ARG_CACHE_TYPE_K)

-ctv, --cache-type-v TYPE KV cache data type for V

(default: f16) (env: LLAMA_ARG_CACHE_TYPE_V)

-dt, --defrag-thold N KV cache defragmentation threshold (DEPRECATED)

(env: LLAMA_ARG_DEFRAG_THOLD)
DEPRECATED in favor of `--load-mode`: force system to keep model in RAM rather than swapping or compressing (env: LLAMA_ARG_MLOCK)
DEPRECATED in favor of `--load-mode`: whether to memory-map model. (if mmap disabled, slower load but may reduce pageouts if not using mlock) (env: LLAMA_ARG_MMAP)

-dio, --direct-io, -ndio, --no-direct-io

(env: LLAMA_ARG_DIO)

-lm, --load-mode MODE model loading mode (default: mmap)

- mmap: memory-map model (if mmap disabled, slower load but may reduce pageouts if not using mlock) - mlock: force system to keep model in RAM rather than swapping or compressing - mmap+mlock: mmap + force system to keep model in RAM rather than swapping or compressing - dio: use DirectIO if available
(env: LLAMA_ARG_LOAD_MODE)
attempt optimizations that help on some NUMA systems - distribute: spread execution evenly over all nodes - isolate: only spawn threads on CPUs on the node that execution started on - numactl: use the CPU map provided by numactl if run without this previously, it is recommended to drop the system page cache before using this see https://github.com/ggml-org/llama.cpp/issues/1437 (env: LLAMA_ARG_NUMA)

-dev, --device <dev1,dev2,..> comma-separated list of devices to use for offloading (none = don't

use --list-devices to see a list of available devices (env: LLAMA_ARG_DEVICE)
print list of available devices and exit

-ot, --override-tensor <tensor name pattern>=<buffer type>,...

(env: LLAMA_ARG_OVERRIDE_TENSOR)

-cmoe, --cpu-moe keep all Mixture of Experts (MoE) weights in the CPU

(env: LLAMA_ARG_CPU_MOE)

-ncmoe, --n-cpu-moe N keep the Mixture of Experts (MoE) weights of the first N layers in the

(env: LLAMA_ARG_N_CPU_MOE)

-ngl, --gpu-layers, --n-gpu-layers N max. number of layers to store in VRAM, either an exact number,

'auto', or 'all' (default: auto)
(env: LLAMA_ARG_N_GPU_LAYERS)

-sm, --split-mode {none,layer,row,tensor}

- none: use one GPU only - layer (default): split layers and KV across GPUs (pipelined) - row: split weight across GPUs by rows (parallelized) - tensor: split weights and KV across GPUs (parallelized, EXPERIMENTAL) (env: LLAMA_ARG_SPLIT_MODE)

-ts, --tensor-split N0,N1,N2,... fraction of the model to offload to each GPU, comma-separated list of

(env: LLAMA_ARG_TENSOR_SPLIT)

-mg, --main-gpu INDEX the GPU to use for the model (with split-mode = none), or for

(env: LLAMA_ARG_MAIN_GPU)

-fit, --fit [on|off] whether to adjust unset arguments to fit in device memory ('on' or

'off', default: 'on')
(env: LLAMA_ARG_FIT)

-fitt, --fit-target MiB0,MiB1,MiB2,...

single value is broadcast across all devices, default: 1024 (env: LLAMA_ARG_FIT_TARGET)

-fitc, --fit-ctx N minimum ctx size that can be set by --fit option, default: 4096

(env: LLAMA_ARG_FIT_CTX)
check model tensor data for invalid values (default: false)
advanced option to override model metadata by key. to specify multiple overrides, either use comma-separated values. types: int, float, bool, str. example: --override-kv tokenizer.ggml.add_bos_token=bool:false,tokenizer.ggml.add_eos_token=bool:false
whether to offload host tensor operations to device (default: true)
path to LoRA adapter (use comma-separated values to load multiple adapters)
path to LoRA adapter with user defined scaling (format: FNAME:SCALE,...) note: use comma-separated values
add a control vector note: use comma-separated values to add multiple control vectors
add a control vector with user defined scaling SCALE note: use comma-separated values (format: FNAME:SCALE,...)
layer range to apply the control vector(s) to, start and end inclusive

-m, --model FNAME model path to load

(env: LLAMA_ARG_MODEL)

-mu, --model-url MODEL_URL model download url (default: unused)

(env: LLAMA_ARG_MODEL_URL)

-dr, --docker-repo [<repo>/]<model>[:quant]

is optional, default to :latest. example: gemma3 (default: unused) (env: LLAMA_ARG_DOCKER_REPO)

-hf, -hfr, --hf-repo <user>/<model>[:quant]

default to Q4_K_M, or falls back to the first file in the repo if Q4_K_M doesn't exist. mmproj is also downloaded automatically if available. to disable, add --no-mmproj example: ggml-org/GLM-4.7-Flash-GGUF:Q4_K_M (default: unused) (env: LLAMA_ARG_HF_REPO)

-hff, --hf-file FILE Hugging Face model file. If specified, it will override the quant in

(env: LLAMA_ARG_HF_FILE)

-hft, --hf-token TOKEN Hugging Face access token (default: value from HF_TOKEN environment

(env: HF_TOKEN)
Log disable
Log to file (env: LLAMA_ARG_LOG_FILE)
Set colored logging ('on', 'off', or 'auto', default: 'auto') 'auto' enables colors when output is to a terminal (env: LLAMA_ARG_LOG_COLORS)

-v, --verbose, --log-verbose Set verbosity level to infinity (i.e. log all messages, useful for

debugging)
Offline mode: forces use of cache, prevents network access (env: LLAMA_ARG_OFFLINE)

-lv, --verbosity, --log-verbosity N Set the verbosity threshold. Messages with a higher verbosity will be

- 0: generic output - 1: error - 2: warning - 3: info - 4: trace (more info) - 5: debug
(default: 3)
(env: LLAMA_ARG_LOG_VERBOSITY)
Enable prefix in log messages (env: LLAMA_ARG_LOG_PREFIX)
Enable timestamps in log messages (env: LLAMA_ARG_LOG_TIMESTAMPS)
KV cache data type for K for the draft model allowed values: f32, f16, bf16, q8_0, q4_0, q4_1, iq4_nl, q5_0, q5_1 (default: f16) (env: LLAMA_ARG_SPEC_DRAFT_CACHE_TYPE_K)
KV cache data type for V for the draft model allowed values: f32, f16, bf16, q8_0, q4_0, q4_1, iq4_nl, q5_0, q5_1 (default: f16) (env: LLAMA_ARG_SPEC_DRAFT_CACHE_TYPE_V)

----- sampling params -----

samplers that will be used for generation in the order, separated by ';' (default: penalties;dry;top_n_sigma;top_k;typ_p;top_p;min_p;xtc;temperature)

-s, --seed SEED RNG seed (default: -1, use random seed for -1)

simplified sequence for samplers that will be used (default: edskypmxt)
ignore end of stream token and continue generating (implies --logit-bias EOS-inf)
temperature (default: 0.80)
top-k sampling (default: 40, 0 = disabled) (env: LLAMA_ARG_TOP_K)
top-p sampling (default: 0.95, 1.0 = disabled)
min-p sampling (default: 0.05, 0.0 = disabled)
top-n-sigma sampling (default: -1.00, -1.0 = disabled)
xtc probability (default: 0.00, 0.0 = disabled)
xtc threshold (default: 0.10, 1.0 = disabled)
locally typical sampling, parameter p (default: 1.00, 1.0 = disabled)
last n tokens to consider for penalize (default: 64, 0 = disabled)
penalize repeat sequence of tokens (default: 1.00, 1.0 = disabled)
repeat alpha presence penalty (default: 0.00, 0.0 = disabled)
repeat alpha frequency penalty (default: 0.00, 0.0 = disabled)
set DRY sampling multiplier (default: 0.00, 0.0 = disabled)
set DRY sampling base value (default: 1.75)
set allowed length for DRY sampling (default: 2)
set DRY penalty for the last n tokens (default: 64, 0 = disable)
add sequence breaker for DRY sampling, clearing out default breakers ('\n', ':', '"', '*') in the process; use "none" to not use any sequence breakers
adaptive-p: select tokens near this probability (valid range 0.0 to 1.0; negative = disabled) (default: -1.00) [(more info)](https://github.com/ggml-org/llama.cpp/pull/17927)
adaptive-p: decay rate for target adaptation over time. lower values are more reactive, higher values are more stable. (valid range 0.0 to 0.99) (default: 0.90)
dynamic temperature range (default: 0.00, 0.0 = disabled)
dynamic temperature exponent (default: 1.00)
use Mirostat sampling. Top K, Nucleus and Locally Typical samplers are ignored if used. (default: 0, 0 = disabled, 1 = Mirostat, 2 = Mirostat 2.0)
Mirostat learning rate, parameter eta (default: 0.10)
Mirostat target entropy, parameter tau (default: 5.00)

-l, --logit-bias TOKEN_ID(+/-)BIAS modifies the likelihood of token appearing in the completion,

or `--logit-bias 15043-1` to decrease likelihood of token ' Hello'
BNF-like grammar to constrain generations (see samples in grammars/ dir)
file to read grammar from

-j, --json-schema SCHEMA JSON schema to constrain generations (https://json-schema.org/), e.g.

`{}` for any JSON object
For schemas w/ external $refs, use --grammar + example/json_schema_to_grammar.py instead

-jf, --json-schema-file FILE File containing a JSON schema to constrain generations

(https://json-schema.org/), e.g. `{}` for any JSON object
For schemas w/ external $refs, use --grammar + example/json_schema_to_grammar.py instead

-bs, --backend-sampling enable backend sampling (experimental) (default: disabled)

(env: LLAMA_ARG_BACKEND_SAMPLING)

----- speculative params -----

Same as --hf-repo, but for the draft model (default: unused) (env: LLAMA_ARG_SPEC_DRAFT_HF_REPO)
number of threads to use during generation (default: same as --threads)
number of threads to use during batch and prompt processing (default: same as --threads-draft)
Draft model CPU affinity mask. Complements cpu-range-draft (default: same as --cpu-mask)
Ranges of CPUs for affinity. Complements --cpu-mask-draft
Use strict CPU placement for draft model (default: same as --cpu-strict)
set draft process/thread priority : 0-normal, 1-medium, 2-high, 3-realtime (default: 0)
Use polling to wait for draft model work (default: same as --poll)
Draft model CPU affinity mask. Complements cpu-range-draft (default: same as --cpu-mask)
Use strict CPU placement for draft model (default: --cpu-strict-draft)
set draft process/thread priority : 0-normal, 1-medium, 2-high, 3-realtime (default: 0)
Use polling to wait for draft model work (default: --poll-draft)
override tensor buffer type for draft model
keep all Mixture of Experts (MoE) weights in the CPU for the draft model (env: LLAMA_ARG_SPEC_DRAFT_CPU_MOE)
keep the Mixture of Experts (MoE) weights of the first N layers in the CPU for the draft model (env: LLAMA_ARG_SPEC_DRAFT_N_CPU_MOE)
number of tokens to draft for speculative decoding (default: 3) (env: LLAMA_ARG_SPEC_DRAFT_N_MAX)
minimum number of draft tokens to use for speculative decoding (default: 0) (env: LLAMA_ARG_SPEC_DRAFT_N_MIN)
speculative decoding split probability (default: 0.10) (env: LLAMA_ARG_SPEC_DRAFT_P_SPLIT)
minimum speculative decoding probability (greedy) (default: 0.00) (env: LLAMA_ARG_SPEC_DRAFT_P_MIN)
offload draft sampling to the backend (default: enabled) (env: LLAMA_ARG_SPEC_DRAFT_BACKEND_SAMPLING)
comma-separated list of devices to use for offloading the draft model (none = don't offload) use --list-devices to see a list of available devices
max. number of draft model layers to store in VRAM, either an exact number, 'auto', or 'all' (default: auto) (env: LLAMA_ARG_N_GPU_LAYERS_DRAFT)
draft model for speculative decoding (default: unused) (env: LLAMA_ARG_SPEC_DRAFT_MODEL)
comma-separated list of types of speculative decoding to use (default: none)
(env: LLAMA_ARG_SPEC_TYPE)
minimum number of ngram tokens to use for ngram-based speculative decoding (default: 48)
maximum number of ngram tokens to use for ngram-based speculative decoding (default: 64)
ngram-mod lookup length (default: 24)
ngram size N for ngram-simple speculative decoding, length of lookup n-gram (default: 12)
ngram size M for ngram-simple speculative decoding, length of draft m-gram (default: 48)
minimum hits for ngram-simple speculative decoding (default: 1)
ngram size N for ngram-map-k speculative decoding, length of lookup n-gram (default: 12)
ngram size M for ngram-map-k speculative decoding, length of draft m-gram (default: 48)
minimum hits for ngram-map-k speculative decoding (default: 1)
ngram size N for ngram-map-k4v speculative decoding, length of lookup n-gram (default: 12)
ngram size M for ngram-map-k4v speculative decoding, length of draft m-gram (default: 48)
minimum hits for ngram-map-k4v speculative decoding (default: 1)
the argument has been removed. use --spec-draft-n-max or --spec-ngram-mod-n-max (env: LLAMA_ARG_DRAFT_MAX)
the argument has been removed. use --spec-draft-n-min or --spec-ngram-mod-n-min (env: LLAMA_ARG_DRAFT_MIN)
the argument has been removed. use the respective --spec-ngram-*-size-n or --spec-ngram-mod-n-match
the argument has been removed. use the respective --spec-ngram-*-size-m
the argument has been removed. use the respective --spec-ngram-*-min-hits

----- example-specific params -----

-lcs, --lookup-cache-static FNAME path to static lookup cache to use for lookup decoding (not updated by

generation)

-lcd, --lookup-cache-dynamic FNAME path to dynamic lookup cache to use for lookup decoding (updated by

generation)

-ctxcp, --ctx-checkpoints, --swa-checkpoints N

32)[(more info)](https://github.com/ggml-org/llama.cpp/pull/15293) (env: LLAMA_ARG_CTX_CHECKPOINTS)

-cms, --checkpoint-min-step N minimum spacing between context checkpoints in tokens (default: 8192,

0 = no minimum)
(env: LLAMA_ARG_CHECKPOINT_MIN_SPACING_NT)

-cram, --cache-ram N set the maximum cache size in MiB (default: 8192, -1 - no limit, 0 -

info)](https://github.com/ggml-org/llama.cpp/pull/16391) (env: LLAMA_ARG_CACHE_RAM)

-kvu, --kv-unified, -no-kvu, --no-kv-unified

enabled if number of slots is auto) (env: LLAMA_ARG_KV_UNIFIED)
save idle slots to the prompt cache on new task, and clear them when using unified KV (default: enabled, requires cache-ram) (env: LLAMA_ARG_CACHE_IDLE_SLOTS)
whether to use context shift on infinite text generation (default: disabled) (env: LLAMA_ARG_CONTEXT_SHIFT)

-r, --reverse-prompt PROMPT halt generation at PROMPT, return control in interactive mode -sp, --special special tokens output enabled (default: false)

whether to perform warmup with an empty run (default: enabled)
use Suffix/Prefix/Middle pattern for infill (instead of Prefix/Suffix/Middle) as some models prefer this. (default: disabled)
pooling type for embeddings, use model default if unspecified (env: LLAMA_ARG_POOLING)

-np, --parallel N number of server slots (default: -1, -1 = auto)

(env: LLAMA_ARG_N_PARALLEL)

-cb, --cont-batching, -nocb, --no-cont-batching

(default: enabled) (env: LLAMA_ARG_CONT_BATCHING)

-mm, --mmproj FILE path to a multimodal projector file. see tools/mtmd/README.md

(env: LLAMA_ARG_MMPROJ)

-mmu, --mmproj-url URL URL to a multimodal projector file. see tools/mtmd/README.md

(env: LLAMA_ARG_MMPROJ_URL)
whether to use multimodal projector file (if available), useful when using -hf (default: enabled) (env: LLAMA_ARG_MMPROJ_AUTO)
whether to enable GPU offloading for multimodal projector (default: enabled) (env: LLAMA_ARG_MMPROJ_OFFLOAD)
minimum number of tokens each image can take, only used by vision models with dynamic resolution (default: read from model) (env: LLAMA_ARG_IMAGE_MIN_TOKENS)
maximum number of tokens each image can take, only used by vision models with dynamic resolution (default: read from model) (env: LLAMA_ARG_IMAGE_MAX_TOKENS)
maximum number of image tokens per batch when encoding images (default: 1024) (env: LLAMA_ARG_MTMD_BATCH_MAX_TOKENS)

-a, --alias STRING set model name aliases, comma-separated (to be used by API)

(env: LLAMA_ARG_ALIAS)
set model tags, comma-separated (informational, not used for routing) (env: LLAMA_ARG_TAGS)
normalisation for embeddings (default: 2) (-1=none, 0=max absolute int16, 1=taxicab, 2=euclidean, >2=p-norm)
ip address to listen, or bind to an UNIX socket if the address ends with .sock (default: 127.0.0.1) (env: LLAMA_ARG_HOST)
port to listen (default: 8080) (env: LLAMA_ARG_PORT)
allow multiple sockets to bind to the same port (default: disabled) (env: LLAMA_ARG_REUSE_PORT)
path to serve static files from (default: ) (env: LLAMA_ARG_STATIC_PATH)
comma-separated list of allowed origins for CORS (default: *) if set to special value 'localhost', reflect the Origin header only if it is localhost (env: LLAMA_ARG_CORS_ORIGINS)
comma-separated list of allowed methods for CORS (default: GET, POST, DELETE, OPTIONS) (env: LLAMA_ARG_CORS_METHODS)
comma-separated list of allowed headers for CORS (default: *) (env: LLAMA_ARG_CORS_HEADERS)
whether to allow credentials for CORS (default: enabled) note: if this is enabled and --cors-origins is set to * (default), the Origin header will be echoed back, and credentials will always be allowed (env: LLAMA_ARG_CORS_CREDENTIALS)
prefix path the server serves from, without the trailing slash (default: ) (env: LLAMA_ARG_API_PREFIX)
JSON that provides default UI settings (overrides UI defaults) (env: LLAMA_ARG_UI_CONFIG)
JSON file that provides default UI settings (overrides UI defaults) (env: LLAMA_ARG_UI_CONFIG_FILE)
experimental: whether to enable MCP CORS proxy - do not enable in untrusted environments (default: disabled) (env: LLAMA_ARG_UI_MCP_PROXY)
experimental: whether to enable built-in tools for AI agents - do not enable in untrusted environments (default: no tools) specify "all" to enable all tools available tools: read_file, file_glob_search, grep_search, exec_shell_command, write_file, edit_file, get_datetime, get_info note: for security reasons, this will limit --cors-origins to localhost by default (env: LLAMA_ARG_TOOLS)
experimental: run tools in a separate runtime environment (default: none, use host environment) available options: 'docker:<image>': spin up a new Docker container and reuse it for all invocations, clean up on server exit 'docker-container:<id>': use an existing Docker container by ID, won't stop on server exit
(env: LLAMA_ARG_TOOLS_RUNTIME)
experimental: path to JSON file with MCP server definitions (Cursor-compatible format) - do not enable in untrusted environments (default: none) note: for security reasons, this will limit --cors-origins to localhost by default (env: LLAMA_ARG_MCP_SERVERS_CONFIG)
experimental: inline JSON with MCP server definitions (Cursor-compatible format) - do not enable in untrusted environments (default: none) note: for security reasons, this will limit --cors-origins to localhost by default (env: LLAMA_ARG_MCP_SERVERS_JSON)

-ag, --agent, -no-ag, --no-agent whether to enable CORS proxy and all built-in tools - do not enable in

note: for security reasons, this will limit --cors-origins to localhost by default (env: LLAMA_ARG_AGENT)
whether to enable the Web UI (default: enabled) (env: LLAMA_ARG_UI)
restrict to only support embedding use case; use only with dedicated embedding models (default: disabled) (env: LLAMA_ARG_EMBEDDINGS)
enable reranking endpoint on server (default: disabled) (env: LLAMA_ARG_RERANKING)
API key to use for authentication, multiple keys can be provided as a comma-separated list (default: none) (env: LLAMA_API_KEY)
path to file containing API keys, one per line; lines starting with a hash are treated as comments (default: none) (env: LLAMA_ARG_API_KEY_FILE)
path to file a PEM-encoded SSL private key (env: LLAMA_ARG_SSL_KEY_FILE)
path to file a PEM-encoded SSL certificate (env: LLAMA_ARG_SSL_CERT_FILE)
sets additional params for the json template parser, must be a valid json object string, e.g. '{"key1":"value1","key2":"value2"}' (env: LLAMA_ARG_CHAT_TEMPLATE_KWARGS)

-to, --timeout N server read/write timeout in seconds (default: 3600)

(env: LLAMA_ARG_TIMEOUT)
server SSE ping interval in seconds (-1 = disabled, default: 30) (env: LLAMA_ARG_SSE_PING_INTERVAL)
number of threads used to process HTTP requests (default: -1) (env: LLAMA_ARG_THREADS_HTTP)
whether to enable prompt caching (default: enabled) (env: LLAMA_ARG_CACHE_PROMPT)
min chunk size to attempt reusing from the cache via KV shifting, requires prompt caching to be enabled (default: 0) [(card)](https://ggml.ai/f0.png) (env: LLAMA_ARG_CACHE_REUSE)
enable prometheus compatible metrics endpoint (default: disabled) (env: LLAMA_ARG_ENDPOINT_METRICS)
enable changing global properties via POST /props (default: disabled) (env: LLAMA_ARG_ENDPOINT_PROPS)
expose slots monitoring endpoint (default: enabled) (env: LLAMA_ARG_ENDPOINT_SLOTS)
path to save slot kv cache (default: disabled)
directory for loading local media files; files can be accessed via file:// URLs using relative paths (default: disabled)
directory containing models for the router server (default: disabled) (env: LLAMA_ARG_MODELS_DIR)
path to INI file containing model presets for the router server (default: disabled) (env: LLAMA_ARG_MODELS_PRESET)
for router server, maximum number of models to load simultaneously (default: 4, 0 = unlimited) (env: LLAMA_ARG_MODELS_MAX)
for router server, whether to automatically load models (default: enabled) (env: LLAMA_ARG_MODELS_AUTOLOAD)
whether to use jinja template engine for chat (default: enabled) (env: LLAMA_ARG_JINJA)
controls whether thought tags are allowed and/or extracted from the response, and in which format they're returned; one of: - none: leaves thoughts unparsed in `message.content` - deepseek: puts thoughts in `message.reasoning_content` - deepseek-legacy: keeps `<think>` tags in `message.content` while also populating `message.reasoning_content` (default: auto) (env: LLAMA_ARG_THINK)

-rea, --reasoning [on|off|auto] Use reasoning/thinking in the chat ('on', 'off', or 'auto', default:

'auto' (detect from template))
(env: LLAMA_ARG_REASONING)
token budget for thinking: -1 for unrestricted, 0 for immediate end, N>0 for token budget (default: -1) (env: LLAMA_ARG_THINK_BUDGET)
message injected before the end-of-thinking tag when reasoning budget is exhausted (default: none) (env: LLAMA_ARG_THINK_BUDGET_MESSAGE)
preserve reasoning trace in the full history, not just the last assistant message (default: template default) compatible with certain templates having 'supports_preserve_reasoning' capability example: https://docs.z.ai/guides/capabilities/thinking-mode#preserved-thinking (env: LLAMA_ARG_REASONING_PRESERVE)
set custom jinja chat template (default: template taken from model's metadata) if suffix/prefix are specified, template will be disabled only commonly used templates are accepted (unless --jinja is set before this flag): list of built-in templates: bailing, bailing-think, bailing2, chatglm3, chatglm4, chatml, command-r, deepseek, deepseek-ocr, deepseek2, deepseek3, exaone-moe, exaone3, exaone4, falcon3, gemma, gigachat, glmedge, gpt-oss, granite, granite-4.0, granite-4.1, grok-2, hunyuan-dense, hunyuan-moe, hunyuan-vl, kimi-k2, llama2, llama2-sys, llama2-sys-bos, llama2-sys-strip, llama3, llama4, megrez, minicpm, mistral-v1, mistral-v3, mistral-v3-tekken, mistral-v7, mistral-v7-tekken, monarch, openchat, orion, pangu-embedded, phi3, phi4, rwkv-world, seed_oss, smolvlm, solar-open, vicuna, vicuna-orca, yandex, zephyr (env: LLAMA_ARG_CHAT_TEMPLATE)
set custom jinja chat template file (default: template taken from model's metadata) if suffix/prefix are specified, template will be disabled only commonly used templates are accepted (unless --jinja is set before this flag): list of built-in templates: bailing, bailing-think, bailing2, chatglm3, chatglm4, chatml, command-r, deepseek, deepseek-ocr, deepseek2, deepseek3, exaone-moe, exaone3, exaone4, falcon3, gemma, gigachat, glmedge, gpt-oss, granite, granite-4.0, granite-4.1, grok-2, hunyuan-dense, hunyuan-moe, hunyuan-vl, kimi-k2, llama2, llama2-sys, llama2-sys-bos, llama2-sys-strip, llama3, llama4, megrez, minicpm, mistral-v1, mistral-v3, mistral-v3-tekken, mistral-v7, mistral-v7-tekken, monarch, openchat, orion, pangu-embedded, phi3, phi4, rwkv-world, seed_oss, smolvlm, solar-open, vicuna, vicuna-orca, yandex, zephyr (env: LLAMA_ARG_CHAT_TEMPLATE_FILE)
force a pure content parser, even if a Jinja template is specified; model will output everything in the content section, including any reasoning and/or tool calls (default: disabled) (env: LLAMA_ARG_SKIP_CHAT_PARSING)
whether to prefill the assistant's response if the last message is an assistant message (default: prefill enabled) when this flag is set, if the last message is an assistant message then it will be treated as a full message and not prefilled
(env: LLAMA_ARG_PREFILL_ASSISTANT)

-sps, --slot-prompt-similarity SIMILARITY

order to use that slot (default: 0.10, 0.0 = disabled)
load LoRA adapters without applying them (apply later via POST /lora-adapters) (default: disabled)
number of seconds of idleness after which the server will sleep (default: -1; -1 = disabled)
Log prompts to directory (auto-created if not present; only used for debugging, default: disabled)
use default EmbeddingGemma model (note: can download weights from the internet)
use default Qwen 2.5 Coder 1.5B (note: can download weights from the internet)
use default Qwen 2.5 Coder 3B (note: can download weights from the internet)
use default Qwen 2.5 Coder 7B (note: can download weights from the internet)
use Qwen 2.5 Coder 7B + 0.5B draft for speculative decoding (note: can download weights from the internet)
use Qwen 2.5 Coder 14B + 0.5B draft for speculative decoding (note: can download weights from the internet)
use default Qwen 3 Coder 30B A3B Instruct (note: can download weights from the internet)
use gpt-oss-20b (note: can download weights from the internet)
use gpt-oss-120b (note: can download weights from the internet)
use Gemma 3 4B QAT (note: can download weights from the internet)
use Gemma 3 12B QAT (note: can download weights from the internet)
enable default speculative decoding config
August 2026 debian