ollama - 💡(How to fix) Fix Vulkan don't used from version 0.30 [1 pull requests]

time=2026-06-09T15:52:25.289+03:00 level=INFO source=routes.go:1919 msg="server config" env="map[CUDA_VISIBLE_DEVICES: GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: LLAMA_ARG_FIT: LLAMA_ARG_FIT_TARGET: NO_PROXY: OLLAMA_CONTEXT_LENGTH:32768 OLLAMA_DEBUG:INFO OLLAMA_DEBUG_LOG_REQUESTS:false OLLAMA_EDITOR: OLLAMA_FLASH_ATTENTION:false OLLAMA_GO_TEMPLATE:true OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_IGPU_ENABLE:1 OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MAX_TRANSFER_STREAMS:4 OLLAMA_MODELS:C:\\Users\\daniel\\.ollama\\models OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NO_CLOUD:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false OLLAMA_VULKAN:true ROCR_VISIBLE_DEVICES:]"
time=2026-06-09T15:52:25.298+03:00 level=INFO source=routes.go:1921 msg="Ollama cloud disabled: false"
time=2026-06-09T15:52:25.302+03:00 level=INFO source=images.go:864 msg="total blobs: 33"
time=2026-06-09T15:52:25.304+03:00 level=INFO source=images.go:871 msg="total unused blobs removed: 0"
time=2026-06-09T15:52:25.305+03:00 level=INFO source=routes.go:1981 msg="Listening on 127.0.0.1:11434 (version 0.30.7)"
time=2026-06-09T15:52:25.305+03:00 level=INFO source=runner.go:60 msg="discovering available GPUs..."
time=2026-06-09T15:52:25.308+03:00 level=INFO source=model_list_cache.go:111 msg="model list cache hydration complete" models=9 failures=0 elapsed=3.1868ms
time=2026-06-09T15:52:25.627+03:00 level=INFO source=model_recommendations.go:177 msg="model recommendations cache sleep scheduled" wait=3h14m52.460711024s consecutive_failures=0
time=2026-06-09T15:52:26.220+03:00 level=INFO source=types.go:50 msg="inference compute" id=cpu library=cpu compute="" name=cpu description=cpu libdirs=ollama driver="" pci_id="" type="" total="63.5 GiB" available="38.4 GiB"
time=2026-06-09T15:52:26.220+03:00 level=INFO source=routes.go:2031 msg="vram-based default context" total_vram="0 B" default_num_ctx=4096
[GIN] 2026/06/09 - 15:52:26 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2026/06/09 - 15:52:26 | 200 |            0s |       127.0.0.1 | GET      "/api/version"
[GIN] 2026/06/09 - 15:52:26 | 200 |      1.1116ms |       127.0.0.1 | GET      "/api/tags"
[GIN] 2026/06/09 - 15:52:36 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2026/06/09 - 15:52:36 | 200 |    210.8473ms |       127.0.0.1 | POST     "/api/show"
[GIN] 2026/06/09 - 15:52:36 | 200 |       2.677ms |       127.0.0.1 | POST     "/api/show"
time=2026-06-09T15:52:37.173+03:00 level=INFO source=sched.go:1148 msg="disabling mmap for llama-server load by default" model=C:\Users\daniel\.ollama\models\blobs\sha256-7121486771cbfe218851513210c40b35dbdee93ab1ef43fe36283c883980f0df reason=cpu
time=2026-06-09T15:52:37.173+03:00 level=INFO source=server.go:109 msg="using llama-server for model" model=C:\Users\daniel\.ollama\models\blobs\sha256-7121486771cbfe218851513210c40b35dbdee93ab1ef43fe36283c883980f0df
time=2026-06-09T15:52:37.178+03:00 level=INFO source=llama_server.go:403 msg="starting llama-server" cmd="C:\\Users\\daniel\\AppData\\Local\\Programs\\Ollama\\lib\\ollama\\llama-server.exe --model C:\\Users\\daniel\\.ollama\\models\\blobs\\sha256-7121486771cbfe218851513210c40b35dbdee93ab1ef43fe36283c883980f0df --port 51346 --host 127.0.0.1 --no-webui --offline -c 32768 -np 1 --log-verbosity 4 --no-log-prefix --no-log-timestamps --no-jinja --chat-template chatml --mmproj C:\\Users\\daniel\\.ollama\\models\\blobs\\sha256-7121486771cbfe218851513210c40b35dbdee93ab1ef43fe36283c883980f0df --no-mmap --flash-attn auto -b 1024 -ub 1024"
time=2026-06-09T15:52:37.181+03:00 level=INFO source=sched.go:613 msg="system memory" total="63.5 GiB" free="38.4 GiB" free_swap="41.3 GiB"
time=2026-06-09T15:52:37.181+03:00 level=INFO source=llama_server.go:886 msg="loading model via llama-server" model=C:\Users\daniel\.ollama\models\blobs\sha256-7121486771cbfe218851513210c40b35dbdee93ab1ef43fe36283c883980f0df
time=2026-06-09T15:52:37.181+03:00 level=INFO source=llama_server.go:1131 msg="waiting for llama-server to start responding"
time=2026-06-09T15:52:37.181+03:00 level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server error"
common_params_print_info: build 1 (6f3a9f3de) with Clang 18.1.8 for Windows AMD64
log_info: verbosity = 4 (adjust with the `-lv N` CLI arg)
device_info:
  - CPU     : Intel(R) Core(TM) Ultra 7 255H (65017 MiB, 39261 MiB free)
system_info: n_threads = 8 (n_threads_batch = 8) / 16 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX_VNNI = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | BMI2 = 1 | LLAMAFILE = 1 | OPENMP = 1 | REPACK = 1 | 
srv          init: using 15 threads for HTTP server
srv          init: The UI is disabled
srv          init: Use --ui/--no-ui (or deprecated --webui/--no-webui) to enable/disable
srv         start: binding port with default address family
srv  llama_server: loading model
srv    load_model: loading model 'C:\Users\daniel\.ollama\models\blobs\sha256-7121486771cbfe218851513210c40b35dbdee93ab1ef43fe36283c883980f0df'
handle_gemma4_clip: detected Ollama-format gemma4 GGUF used as mmproj; translating
compat tensor transform: op=F16->F32 promote tensor=v.patch_embd.weight bytes=3538944 duration_ms=4.025 total_ops=1 total_bytes=3538944 total_ms=4.025
time=2026-06-09T15:52:37.433+03:00 level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server loading model"
srv    load_model: [mtmd] estimated worst-case memory usage of mmproj is 1279.96 MiB
common_init_result: fitting params to device memory ...
common_init_result: (for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on)
common_params_fit_impl: getting device memory data for initial parameters:
common_memory_breakdown_print: | memory breakdown [MiB] | total   free    self   model   context   compute    unaccounted |
common_memory_breakdown_print: |   - Host               |                 8961 =  7661 +    1040 +     260                |
common_memory_breakdown_print: |   - CPU_REPACK         |                 8917 =  8917 +       0 +       0                |
common_params_fit_impl: projected to use 8961 MiB of host memory vs. 65017 MiB of total host memory
common_params_fit_impl: will leave 56055 >= 2303 MiB of system memory, no changes needed
common_fit_params: successfully fit params to free device memory
common_fit_params: fitting params to free memory took -0.37 seconds
handle_gemma4: detected Ollama-format gemma4 GGUF; applying compatibility fixes
llama_model_loader: loaded meta data with 52 key-value pairs and 658 tensors from C:\Users\daniel\.ollama\models\blobs\sha256-7121486771cbfe218851513210c40b35dbdee93ab1ef43fe36283c883980f0df (version GGUF V3 (latest))
llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
llama_model_loader: - kv   0:                gemma4.attention.head_count u32              = 16
llama_model_loader: - kv   1:             gemma4.attention.head_count_kv arr[i32,30]      = [8, 8, 8, 8, 8, 2, 8, 8, 8, 8, 8, 2, ...
llama_model_loader: - kv   2:                gemma4.attention.key_length u32              = 512
llama_model_loader: - kv   3:            gemma4.attention.key_length_swa u32              = 256
llama_model_loader: - kv   4:    gemma4.attention.layer_norm_rms_epsilon f32              = 0.000001
llama_model_loader: - kv   5:          gemma4.attention.shared_kv_layers u32              = 0
llama_model_loader: - kv   6:            gemma4.attention.sliding_window u32              = 1024
llama_model_loader: - kv   7:    gemma4.attention.sliding_window_pattern arr[bool,30]     = [true, true, true, true, true, false,...
llama_model_loader: - kv   8:              gemma4.attention.value_length u32              = 512
llama_model_loader: - kv   9:          gemma4.attention.value_length_swa u32              = 256
llama_model_loader: - kv  10:                         gemma4.block_count u32              = 30
llama_model_loader: - kv  11:                      gemma4.context_length u32              = 262144
llama_model_loader: - kv  12:                    gemma4.embedding_length u32              = 2816
llama_model_loader: - kv  13:    gemma4.embedding_length_per_layer_input u32              = 0
llama_model_loader: - kv  14:                        gemma4.expert_count u32              = 128
llama_model_loader: - kv  15:          gemma4.expert_feed_forward_length u32              = 704
llama_model_loader: - kv  16:                   gemma4.expert_used_count u32              = 8
llama_model_loader: - kv  17:                 gemma4.feed_forward_length u32              = 2112
llama_model_loader: - kv  18:             gemma4.final_logit_softcapping f32              = 30.000000
llama_model_loader: - kv  19:                gemma4.rope.dimension_count u32              = 512
llama_model_loader: - kv  20:            gemma4.rope.dimension_count_swa u32              = 256
llama_model_loader: - kv  21:                      gemma4.rope.freq_base f32              = 1000000.000000
llama_model_loader: - kv  22:                  gemma4.rope.freq_base_swa f32              = 10000.000000
llama_model_loader: - kv  23:         gemma4.vision.attention.head_count u32              = 16
llama_model_loader: - kv  24: gemma4.vision.attention.layer_norm_epsilon f32              = 0.000001
llama_model_loader: - kv  25:                  gemma4.vision.block_count u32              = 27
llama_model_loader: - kv  26:             gemma4.vision.embedding_length u32              = 1152
llama_model_loader: - kv  27:          gemma4.vision.feed_forward_length u32              = 4304
llama_model_loader: - kv  28:                 gemma4.vision.num_channels u32              = 3
llama_model_loader: - kv  29:                   gemma4.vision.patch_size u32              = 16
llama_model_loader: - kv  30:       gemma4.vision.projector.scale_factor u32              = 3
llama_model_loader: - kv  31:                       general.architecture str              = gemma4
llama_model_loader: - kv  32:                          general.file_type u32              = 15
llama_model_loader: - kv  33:                    general.parameter_count u64              = 25805936462
llama_model_loader: - kv  34:               general.quantization_version u32              = 2
llama_model_loader: - kv  35:               tokenizer.ggml.add_bos_token bool             = false
llama_model_loader: - kv  36:               tokenizer.ggml.add_eos_token bool             = false
llama_model_loader: - kv  37:              tokenizer.ggml.add_mask_token bool             = false
llama_model_loader: - kv  38:           tokenizer.ggml.add_padding_token bool             = false
llama_model_loader: - kv  39:           tokenizer.ggml.add_unknown_token bool             = false
llama_model_loader: - kv  40:                tokenizer.ggml.bos_token_id u32              = 2
llama_model_loader: - kv  41:                tokenizer.ggml.eos_token_id u32              = 1
llama_model_loader: - kv  42:               tokenizer.ggml.eos_token_ids arr[i32,3]       = [1, 106, 50]
llama_model_loader: - kv  43:               tokenizer.ggml.mask_token_id u32              = 4
llama_model_loader: - kv  44:                      tokenizer.ggml.merges arr[str,514906]  = ["\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n \n", ...
llama_model_loader: - kv  45:            tokenizer.ggml.padding_token_id u32              = 0
llama_model_loader: - kv  46:                         tokenizer.ggml.pre str              = gemma4
llama_model_loader: - kv  47:                      tokenizer.ggml.scores arr[f32,262144]  = [0.000000, 1.000000, 2.000000, 3.0000...
llama_model_loader: - kv  48:                  tokenizer.ggml.token_type arr[i32,262144]  = [3, 3, 3, 3, 3, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv  49:                      tokenizer.ggml.tokens arr[str,262144]  = ["<pad>", "<eos>", "<bos>", "<unk>", ...
llama_model_loader: - kv  50:            tokenizer.ggml.unknown_token_id u32              = 3
llama_model_loader: - kv  51:                       tokenizer.ggml.model str              = gemma4
llama_model_loader: - type  f32:  392 tensors
llama_model_loader: - type q5_0:   32 tensors
llama_model_loader: - type q8_0:   28 tensors
llama_model_loader: - type q4_K:  193 tensors
llama_model_loader: - type q6_K:   13 tensors
print_info: file format = GGUF V3 (latest)
print_info: file type   = Q4_K - Medium
print_info: file size   = 15.63 GiB (5.32 BPW) 
load: override 'tokenizer.ggml.add_bos_token' to 'true' for Gemma4
load: 0 unused tokens
load: control-looking token:    212 '</s>' was not control-type; this is probably a bug in the model. its type will be overridden
load: printing all EOG tokens:
load:   - 1 ('<eos>')
load:   - 50 ('<|tool_response>')
load:   - 106 ('<turn|>')
load:   - 212 ('</s>')
load: special_eog_ids contains '<|tool_response>', removing '</s>' token from EOG list
load: special tokens cache size = 24
load: token to piece cache size = 1.9458 MB
print_info: arch                  = gemma4
print_info: vocab_only            = 0
print_info: no_alloc              = 0
print_info: n_ctx_train           = 262144
print_info: n_embd                = 2816
print_info: n_embd_inp            = 2816
print_info: n_layer               = 30
print_info: n_head                = 16
print_info: n_head_kv             = [8, 8, 8, 8, 8, 2, 8, 8, 8, 8, 8, 2, 8, 8, 8, 8, 8, 2, 8, 8, 8, 8, 8, 2, 8, 8, 8, 8, 8, 2]
print_info: n_rot                 = 512
print_info: n_swa                 = 1024
print_info: is_swa_any            = 1
print_info: n_embd_head_k         = 512
print_info: n_embd_head_v         = 512
print_info: n_gqa                 = [2, 2, 2, 2, 2, 8, 2, 2, 2, 2, 2, 8, 2, 2, 2, 2, 2, 8, 2, 2, 2, 2, 2, 8, 2, 2, 2, 2, 2, 8]
print_info: n_embd_k_gqa          = [2048, 2048, 2048, 2048, 2048, 1024, 2048, 2048, 2048, 2048, 2048, 1024, 2048, 2048, 2048, 2048, 2048, 1024, 2048, 2048, 2048, 2048, 2048, 1024, 2048, 2048, 2048, 2048, 2048, 1024]
print_info: n_embd_v_gqa          = [2048, 2048, 2048, 2048, 2048, 1024, 2048, 2048, 2048, 2048, 2048, 1024, 2048, 2048, 2048, 2048, 2048, 1024, 2048, 2048, 2048, 2048, 2048, 1024, 2048, 2048, 2048, 2048, 2048, 1024]
print_info: f_norm_eps            = 0.0e+00
print_info: f_norm_rms_eps        = 1.0e-06
print_info: f_clamp_kqv           = 0.0e+00
print_info: f_max_alibi_bias      = 0.0e+00
print_info: f_logit_scale         = 0.0e+00
print_info: f_attn_scale          = 1.0e+00
print_info: f_attn_value_scale    = 0.0000
print_info: n_ff                  = 2112
print_info: n_expert              = 128
print_info: n_expert_used         = 8
print_info: n_expert_groups       = 0
print_info: n_group_used          = 0
print_info: causal attn           = 1
print_info: pooling type          = -1
print_info: rope type             = 2
print_info: rope scaling          = linear
print_info: freq_base_train       = 1000000.0
print_info: freq_scale_train      = 1
print_info: freq_base_swa         = 10000.0
print_info: freq_scale_swa        = 1
print_info: n_embd_head_k_swa     = 256
print_info: n_embd_head_v_swa     = 256
print_info: n_rot_swa             = 256
print_info: n_ctx_orig_yarn       = 262144
print_info: rope_yarn_log_mul     = 0.0000
print_info: rope_finetuned        = unknown
print_info: model type            = 26B.A4B
print_info: model params          = 25.23 B
print_info: general.name          = n/a
print_info: vocab type            = BPE
print_info: n_vocab               = 262144
print_info: n_merges              = 514906
print_info: BOS token             = 2 '<bos>'
print_info: EOS token             = 1 '<eos>'
print_info: UNK token             = 3 '<unk>'
print_info: PAD token             = 0 '<pad>'
print_info: MASK token            = 4 '<mask>'
print_info: LF token              = 107 '
'
print_info: EOG token             = 1 '<eos>'
print_info: EOG token             = 50 '<|tool_response>'
print_info: EOG token             = 106 '<turn|>'
print_info: max token length      = 93
load_tensors: loading model tensors, this can take a while... (mmap = false, direct_io = false)
load_tensors:          CPU model buffer size =  7661.57 MiB
load_tensors:   CPU_REPACK model buffer size =  8917.35 MiB
....................................................................
common_init_result: added <eos> logit bias = -inf
common_init_result: added <|tool_response> logit bias = -inf
common_init_result: added <turn|> logit bias = -inf
llama_context: constructing llama_context
llama_context: n_seq_max     = 1
llama_context: n_ctx         = 32768
llama_context: n_ctx_seq     = 32768
llama_context: n_batch       = 1024
llama_context: n_ubatch      = 1024
llama_context: causal_attn   = 1
llama_context: flash_attn    = auto
llama_context: kv_unified    = false
llama_context: freq_base     = 1000000.0
llama_context: freq_scale    = 1
llama_context: n_rs_seq      = 0
llama_context: n_outputs_max = 1
llama_context: n_ctx_seq (32768) < n_ctx_train (262144) -- the full capacity of the model will not be utilized
llama_context:        CPU  output buffer size =     1.00 MiB
llama_kv_cache_iswa: creating non-SWA KV cache, size = 32768 cells
llama_kv_cache:        CPU KV buffer size =   640.00 MiB
llama_kv_cache: size =  640.00 MiB ( 32768 cells,   5 layers,  1/1 seqs), K (f16):  320.00 MiB, V (f16):  320.00 MiB
llama_kv_cache: attn_rot_k = 0, n_embd_head_k_all = 512
llama_kv_cache: attn_rot_v = 0, n_embd_head_k_all = 512
llama_kv_cache_iswa: creating     SWA KV cache, size = 2048 cells
llama_kv_cache:        CPU KV buffer size =   400.00 MiB
llama_kv_cache: size =  400.00 MiB (  2048 cells,  25 layers,  1/1 seqs), K (f16):  200.00 MiB, V (f16):  200.00 MiB
llama_kv_cache: attn_rot_k = 0, n_embd_head_k_all = 256
llama_kv_cache: attn_rot_v = 0, n_embd_head_k_all = 256
sched_reserve: reserving ...
sched_reserve: Flash Attention was auto, set to enabled
sched_reserve: resolving fused Gated Delta Net support:
sched_reserve: fused Gated Delta Net (autoregressive) enabled
sched_reserve: fused Gated Delta Net (chunked) enabled
sched_reserve:        CPU compute buffer size =   260.04 MiB
sched_reserve: graph nodes  = 2645
sched_reserve: graph splits = 1
sched_reserve: reserve took 3.87 ms, sched copies = 1
common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable)
handle_gemma4_clip: detected Ollama-format gemma4 GGUF used as mmproj; translating
clip_model_loader: model name:   
clip_model_loader: description:  
clip_model_loader: GGUF version: 3
clip_model_loader: alignment:    32
clip_model_loader: n_tensors:    1014
clip_model_loader: n_kv:         64

clip_model_loader: has vision encoder
clip_ctx: CLIP using CPU backend
load_hparams: projector:          gemma4v
load_hparams: n_embd:             1152
load_hparams: n_head:             16
load_hparams: n_ff:               4304
load_hparams: n_layer:            27
load_hparams: ffn_op:             gelu_quick
load_hparams: projection_dim:     2816

--- vision hparams ---
load_hparams: image_size:         224
load_hparams: patch_size:         16
load_hparams: has_llava_proj:     0
load_hparams: minicpmv_version:   0
load_hparams: n_merge:            3
load_hparams: n_wa_pattern: 0
load_hparams: image_min_pixels:   92160
load_hparams: image_max_pixels:   645120

load_hparams: model size:         17140.88 MiB
load_hparams: metadata size:      0.36 MiB
compat tensor transform: op=F16->F32 promote tensor=v.patch_embd.weight bytes=3538944 duration_ms=4.212 total_ops=2 total_bytes=7077888 total_ms=8.237
warmup: warmup with image size = 768 x 768
reserve_compute_meta:        CPU compute buffer size =   140.50 MiB
reserve_compute_meta: graph splits = 1, nodes = 1569
warmup: flash attention is enabled
srv    load_model: loaded multimodal model, 'C:\Users\daniel\.ollama\models\blobs\sha256-7121486771cbfe218851513210c40b35dbdee93ab1ef43fe36283c883980f0df'
srv    load_model: initializing slots, n_slots = 1
common_speculative_init: no implementations specified for speculative decoding
slot   load_model: id  0 | task -1 | new slot, n_ctx = 32768
srv    load_model: prompt cache is enabled, size limit: 8192 MiB
srv    load_model: use `--cache-ram 0` to disable the prompt cache
srv    load_model: for more info see https://github.com/ggml-org/llama.cpp/pull/16391
srv    load_model: context checkpoints enabled, max = 32, min spacing = 256
srv          init: --cache-idle-slots requires --kv-unified, disabling
init: chat template, example_format: '<|im_start|>system
You are a helpful assistant<|im_end|>
<|im_start|>user
Hello<|im_end|>
<|im_start|>assistant
Hi there<|im_end|>
<|im_start|>user
How are you?<|im_end|>
<|im_start|>assistant
'
srv          init: init: chat template, thinking = 0
srv  llama_server: model loaded
srv  llama_server: server is listening on http://127.0.0.1:51346
srv  update_slots: all slots are idle
time=2026-06-09T15:52:48.742+03:00 level=INFO source=llama_server.go:1198 msg="llama-server started in 11.56 seconds"
time=2026-06-09T15:52:48.791+03:00 level=INFO source=images.go:354 msg="template selection" model=registry.ollama.ai/library/gemma4:26b selected=renderer_parser renderer=gemma4 parser=gemma4 go_template=null chat_template=null harmony=null renderer_parser="[completion vision tools thinking]"
time=2026-06-09T15:52:48.791+03:00 level=INFO source=sched.go:729 msg="loaded runners" count=1
time=2026-06-09T15:52:48.791+03:00 level=INFO source=llama_server.go:1131 msg="waiting for llama-server to start responding"
time=2026-06-09T15:52:48.791+03:00 level=INFO source=llama_server.go:1198 msg="llama-server started in 11.61 seconds"
[GIN] 2026/06/09 - 15:52:48 | 200 |     12.00254s |       127.0.0.1 | POST     "/api/generate"