The vision encoder sits idle in VRAM on every text request. With
LLAMA_MMPROJ_ONDEMAND=1 the server keeps the encoder in RAM (released
after load) and brings it into VRAM just-in-time before mtmd_batch_encode,
releasing it again afterwards - so the encoder's VRAM is free for KV /
expert cache on the common text-only path.
When the encoder does not fit (that VRAM has been claimed), evict the LLM
backbone weights first: they are not needed while the encoder runs (the
decode that uses them happens afterwards), their host shadow is read-only
(cheap free, no D2H), and the KV / prompt cache is left untouched. Order
on the way back matters: free the encoder before re-uploading the weights
so the peak stays within VRAM.
Also wire the encoder release/restore into the VRAM arbiter: vram_go_cold
releases it; the just-in-time encode-path restore replaces the previous
restore in vram_ensure_warm.
Assisted-by: Claude
With LLAMA_SLEEP_EVICT_KV the KV cache device buffers are freed when a model
goes cold. update_slots() touches the KV before the decode-time vram_ensure_warm
(e.g. SWA models create a checkpoint that reads the KV via ggml_backend_tensor_get),
so the KV must already be resident by then. Move the restore to the sleep-wake
handler (handle_sleeping_state(false)), which runs before update_slots, fixing a
GGML_ASSERT(buffer) crash on iSWA models (gemma) when KV eviction is enabled.
Assisted-by: Claude
Before uploading a model's weights, acquire the shared VRAM token (ring the
doorbell so any resident model releases first). Previously load uploaded weights
to VRAM before the arbiter was active, so loading a large model while another
(e.g. the warm 4B task model) held VRAM could exceed the budget and OOM.
- add vram_arena_open() (idempotent flock/doorbell setup) and
vram_acquire_for_load(), called from load_model() before
common_init_from_params().
- a coordinated load now stays warm holding the token and serves its first
request without a re-warm (drop the init-time go_cold cold-start).
Assisted-by: Claude
The request handler only calls wait_until_no_sleep() (which wakes a server out
of its sleeping state) when sleep_idle_seconds >= 0. But the VRAM arbiter's
cross-process doorbell can put a server to sleep even when idle-sleep is
disabled, so without this a doorbell-slept server would never wake and requests
to it would hang until timeout.
Do not bypass the wake path when LLAMA_SLEEP_VRAM_ONLY is set, so the arbiter no
longer depends on --sleep-idle-seconds being configured.
Assisted-by: Claude
Extend on-demand device residency to the KV cache so that when a model's KV
plus another model would not fit in VRAM, the KV can also be evicted to a host
shadow (D2H on release, H2D on restore) instead of only the weights.
- llama_memory_i: add release_device_buffers()/restore_device_buffers()
(default no-op). Implemented in llama_kv_cache (D2H shadow of the live
ctxs_bufs, freed and reallocated like the weights); llama_memory_hybrid and
llama_kv_cache_iswa delegate to their child caches.
- llama_context::release_device(evict_kv): also evict the memory's device
buffers when requested; restore_device() rebuilds them. Public API
llama_context_release_device gains an evict_kv flag.
- server: LLAMA_SLEEP_EVICT_KV=1 enables it. Off by default (weights-only),
since the KV shadow adds a D2H/H2D copy of the live cache each cycle.
Validated on RX 580 (Vulkan), 4B @ 32k ctx: weights-only cold VRAM 1750 MB
(KV stays); weights+KV cold VRAM 726 MB (KV freed, ~1 GB reclaimed). KV
survives the round-trip: prompt cache reused after the cycle (prompt_n 4 vs
42), correct output.
Assisted-by: Claude
Add release/restore of a model's GPU weight buffers (keeping a host shadow
and the KV cache) so several always-loaded llama-server processes can
time-share a single GPU without reloading or losing the prompt cache.
- llama-model: release_device_weights()/restore_device_weights() capture a
compact host shadow (stable iteration order, view-skipping) and free then
realloc the device weight buffers; weights_resident() query.
- llama-context: release_device()/restore_device() wrappers; decode() auto-
restores; public C API llama_context_release_device/restore_device.
- server: LLAMA_SLEEP_VRAM_ONLY makes idle-sleep release only the VRAM weights
(not a full unload/reload). A cross-process flock token in LLAMA_VRAM_ARENA
enforces "resident iff holds token"; an inotify doorbell forces the holder
to release on contention. The warden thread only touches the task queue, so
releases run on the loop thread and never race a decode.
Validated on RX 580 (Vulkan): two models share 8GB, never both resident,
correct output under contention, KV cache preserved (no re-prefill).
Assisted-by: Claude
* server : preserve context checkpoints across slot save/restore
Append the checkpoints after the packed server_tokens payload added in #26640
and count them in n_written / n_read, so a restored slot can still roll back to
a checkpoint instead of re-processing the whole prompt.
* server : drop draft checkpoint data that does not match the draft context
Restoring a slot saved with a different draft KV cache type aborted in
load_dft(). Test-load one draft checkpoint on restore and drop the draft
data if it does not fit, instead of crashing. Adds a regression test.
Co-authored-by: Igor Okulist <okigan@gmail.com>
* server : harden the checkpoint appendix of slot save files
Bound each blob size by the bytes left in the file before allocating, open the
file with UTF-8 paths on Windows like the llama state payload, fall back to full
prompt re-processing when a checkpoint restored from a slot file fails to load,
and replace the 1024 count cap by keeping the last n_ctx_checkpoints while reading.
* server : report an incomplete checkpoint appendix as a failed slot save
Return an error to the client when the appendix cannot be written, like a
failed payload write, and make the oversized-blob test declare a size that
cannot be allocated, so an unbounded allocation fails the test.
* server : reject an empty target state in the checkpoint appendix
A saved checkpoint always holds a target state, an empty blob would roll back
without restoring anything. Also log with the slot id, and load the draft test
model from the HF cache instead of a second download.
* common : return bool from checkpoint load_tgt / load_dft
A checkpoint restored from a slot file falls back to full prompt re-processing
when it fails to load, a checkpoint created in memory still aborts.
---------
Co-authored-by: Igor Okulist <okigan@gmail.com>
* model : add LiquidAI/d1-omni-600M decision model
Assisted-by: Claude Opus 5.5
* mtmd : keep conformer GLU sigmoid on CUDA
Assisted-by: Claude Opus 5.5
* server : take d1omni audio through images and input_audio, scope memory-less lfm2 to non-causal
Assisted-by: Claude Opus 5.5
* common : rename decision type d1omni to lfm2-d1-omni, server : make images an alias of files
Assisted-by: Claude Opus 5.5
* model : add LiquidAI/d1-3b decision model
mtmd : read LFM2 image resize algo from GGUF
Assisted-by: Claude Opus 5.5
* common : rename decision type d1 to lfm2-d1
Assisted-by: Claude Opus 5.5
* server: support vision input for Clef
* move input_attn_causal to private
* extend old server_batch::embd
* server_batch::token::pos to multi dim
* nits
* fix abort
* fix img tokens cap
* fix yield_to_queue mutate data
* server: reject partial media truncation
* server: keep only the keep_first fix
Drop the mtmd test helper change, which no longer builds since
clip_image_f32_batch stores its entries by value, and drop the
vision test: no test fixture reaches a cut between two adjacent
media chunks with a reused cache (tinygemma3 uses SWA and wraps
images in text tokens, tinyopenjev and small-test are recurrent),
so the test passed or failed independently of the fix.
---------
Co-authored-by: Pascal <admin@serveurperso.com>
* log, server: make router child lines carry their own colors
The logger writes the color reset after the trailing newline, so the
reset opens the next line. On the shared pipe of a router child it lands
in front of the next state command, which the router then misses, and
the line break that works around it shows up as an empty log line on
every progress update.
The reset now goes before the trailing newlines, so every line is self
contained and the command goes back to its plain framing. The router
passes its effective color setting to its children, whose output ends
up in its terminal, and leaves that option out when comparing presets
on reload.
* log: enable ANSI colors on the Windows console
A Windows console renders ANSI sequences only in virtual terminal mode,
which nothing turns on for the logger, so llama-server prints raw escape
codes on the Windows 10 console while llama-cli, whose console code
enables it, shows colors. The logger now enables virtual terminal mode
on stdout and stderr when it turns colors on, and keeps colors off when
a console cannot render them. Pipes and files take the sequences as is.
* server: separate the router child commands from its logs
The child sent its state commands on the same pipe as its logs, so the
router had to pick them out of the log stream by a line prefix, and any
unterminated write in front of a command made the router miss it. This
resolves the TODO at the spawn that called for splitting stdout and
stderr.
The child now keeps stdout for the commands and points everything else
written to stdout at stderr, before anything is written. The router
reads both pipes, handles the commands from stdout and forwards stderr
as the log, and warns about any other line on the command pipe.
* server: address review from ngxson
The single server_child is now created first in the entry point and its
constructor keeps stdout for the commands, so the stream is a member of
the instance instead of a static, and init() is gone. The instance is
passed down to the server, while the CLI entry point creates its own.
* Update tools/server/server.cpp
---------
Co-authored-by: Xuan-Son Nguyen <thichthat@gmail.com>
* Make the drafter probabilistic and the target verify by rejection sampling
* Drop stale spec_draft_q before drafting
* Fallback to argmax sampling for grammar-constrained requests and adding flag for enabling probabilistic draft sampling. Default flag value is greedy.
* Support grammar-constrained requests in rejection sampling
* Fix - renormalize distribution after masking
* copy rng on sampler copy and re-accept drafted tokens on replay
* Fix draft sampler sharing the target's rng stream
* Simplify the rejection sampler's inputs and move replay to the server
* Truncate the draft candidates along with the draft
---------
Co-authored-by: praneshgo <227579474+praneshgo@users.noreply.github.com>
Co-authored-by: Pranesh Gonegandla <pgonegandla@nvidia.com>
* init conversion
* convert: ok
* model loaded
* add server code
* improve conversion script
* support shared prompt prefix
* add docs, imorove UX a bit
* add vision support
* add openjev tiny model for testing
* add dev docs
* support lev & kev
* clean up
* fix lev noul
* fix py lint
* nits docs
* clarify about not supporting date_facts
With --path or --no-ui, /sw.js returned 404, and a 404 does not remove a service worker, so browsers kept showing the cached built-in UI. Serve a worker that unregisters itself, clears its caches and reloads open tabs. A sw.js in the --path folder is still served first.
Assisted-by: Claude Opus 5.5
* server : support multimodal input for /v1/embeddings (Qwen3-VL-Embedding)
Accept the OpenAI-style wrapped content array format for multimodal
embedding requests. Each {"content": [...]} object is one input that
produces one embedding; text parts are concatenated and image_url parts
are decoded via handle_media then spliced with process_mtmd_prompt.
The legacy formats (plain string, token arrays, mixed arrays, and the
{prompt_string, multimodal_data} object) continue to work unchanged via
tokenize_input_prompts. Bare content arrays (the unwrapped shape) are
rejected with a migration message.
Also disables KV prefix reuse for stateless embedding/rerank tasks so
that repeated inputs do not incorrectly share cached KV across requests.
Assisted-by: Opencode Qwen3.8 27B
* clean up comments and docs
* refactor
* add tests
* support video and audio inp
---------
Co-authored-by: timothywang21 <timothywang21@users.noreply.github.com>
Co-authored-by: Xuan Son Nguyen <son@huggingface.co>
* adapt common
* add common_batch
* wip
* wip: spec
* cont
* common_speculative_process
* server_batch to use common_batch
* rm some stale calls
Assisted-by: Claude Fable 5.1
* migrate mtmd
* handle imrope, handle return val of add()/add_embd()
* add spec zeros vector
* add warning on zero fill path
* server : allow splitting RANK pooling for causal LLM rerankers
Rerank models fall into two categories: bidirectional cross-encoders
(BERT, etc.) that require all tokens in a single physical batch, and
causal LLMs repurposed as rerankers (Qwen3, Qwen3-VL) that can use
chunked prefill like any other decoder.
Previously the server rejected all RANK-pooling inputs larger than
n_ubatch, and the graph builder hardcoded QWEN3/QWEN3VL arch checks to
determine last-token pooling. This broke long-document and multimodal
reranking for causal models.
Fix: expose llama_get_causal_attn(ctx) so the server can check the
effective runtime attention type (reflecting any --attention override
or set_causal_attn call). Also expose llama_model_is_causal(model)
for querying the static architectural property from GGUF metadata.
can_split() now permits chunked prefill for RANK pooling when the
context is causal. The graph builder's inline arch check is replaced
with the same cparams.causal_attn predicate, removing the duplication.
Assisted-by: Opencode/Qwen3.8-27B
* remove unused llama_model_is_causal, fix whitespace
Assisted-by: opencode
---------
Co-authored-by: timothywang21 <timothywang21@users.noreply.github.com>
- register --rpc unconditionally and call llama_supports_rpc() only from its handler
- print server "initialization ..." log after args are parsed
Assisted-by: pi:llama.cpp/MiMo-V2.6-Flash-RL
The OpenAI chat completions API specifies content part type "video_url"
with a {"url": ...} object, and clients typically send data: URIs
(e.g. data:video/mp4;base64,...). The llama-server only accepted the
non-standard "input_video" type and rejected data: URIs for video
(accept_base64_uri=false), so any OpenAI-conformant client failed with
"unsupported content[].type" or "Invalid uri format".
- accept "video_url" as an alias of "input_video"
- read the media object from whichever key was used
- allow data: URIs for video (data:video/*), as already done for images
* server: route every model load through the queue
A model loaded by the fast path has no queue entry, so tick() evicts
it at its LOADED transition before its own request is proxied. Every
load now joins the queue, whose entry protects the model until its
waiters leave.
* server: do not admit requests into a stopping model
A request for a model that is being stopped still sees it LOADED and
is proxied into the dying child. Such a request now joins the queue
and is served by the next instance. The stopping mark is cleared
under the same lock that sets UNLOADED, so no request can see a
model that is neither stopping nor unloaded while its child is gone.
Allow configuring --temp, --top-p, --min-p, --repeat-penalty,
--presence-penalty and --frequency-penalty via LLAMA_ARG_* so
llama-server can be fully controlled from an EnvironmentFile
(e.g. systemd on Debian).
Use `llama-gen-docs` to regenerate the readme files.
In router mode, authentication belongs to the router. unset_reserved_args()
already unset LLAMA_API_KEY, but did not unset LLAMA_ARG_API_KEY_FILE.
When --api-key-file was passed, children re-validated against file keys only,
causing clients using --api-key to 401 on chat completions (#28820).
In addition, router internal calls without auth headers (such as
POST /v1/streams/lookup and DELETE /v1/stream) were silently rejected with 401.
Unset LLAMA_ARG_API_KEY_FILE in unset_reserved_args() so no API keys reach
child instances. This keeps keys out of child argv, ensures all keys the router
accepts work end-to-end, and prevents router internal stream calls from 401ing.
Fixes#28820
* server-models : show source per model in log
- Show [source] tag (preset/models_dir/cache) per model instead of cryptic * marker
- Show HF hub cache path in the 'Loaded cached model presets' log
- Add hf_cache::get_cache_dir() public accessor
Assisted-by: pi:llama.cpp/Qwen3.8-27B
* cont : pad log
This commit updates cmake to use PROJECT_SOURCE_DIR instead of CMAKE_SOURCE_DIR for paths in function calls.
The motivation for this is that when using add_subdirectory,
CMAKE_SOURCE_DIR is fixed to the top-level projects source directory,
that is the caller of add_subdirectory and not the llama.cpp root
which means that common/common.h header will not be resolved.
Refs: https://github.com/ggml-org/llama.cpp/pull/28091#issuecomment-5636106377
This commit removes the precompiled headers that I added in Commit
3bcfeb700 ("cmake : add PCH and unity build to improve build times
(#28091)").
The motivation for this is that this looked good when developing this
but has caused multiple issues that I had taken into consideration and
we have decided to remove it and only keep the unity builds from the
above commit.
Refs: https://github.com/ggml-org/llama.cpp/pull/28882#issuecomment-5662272126