ggml_cuda_init: found 1 CUDA devices (Total VRAM: 8187 MiB): Device 0: NVIDIA GeForce RTX 4060 Laptop GPU, compute capability 8.9, VMM: yes, VRAM: 8187 MiB main: n_parallel is set to auto, using n_parallel = 4 and kv_unified = true build: 8420 (7f2cbd9a4) with Clang 19.1.5 for Windows x86_64 system info: n_threads = 12, n_threads_batch = 16, total_threads = 16 system_info: n_threads = 12 (n_threads_batch = 16) / 16 | CUDA : ARCHS = 750,800,860,890,1200,1210 | USE_GRAPHS = 1 | PEER_MAX_BATCH_SIZE = 128 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX_VNNI = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | BMI2 = 1 | LLAMAFILE = 1 | OPENMP = 1 | REPACK = 1 | Running without SSL init: using 15 threads for HTTP server start: binding port with default address family main: loading model srv load_model: loading model 'models\Qwen3.5-35B-A3B-heretic-Opus-4.6-Distilled.i1-Q4_K_M.gguf' common_init_result: fitting params to device memory, for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on llama_params_fit_impl: projected to use 6074 MiB of device memory vs. 6854 MiB of free device memory llama_params_fit_impl: cannot meet free memory target of 1024 MiB, need to reduce device memory by 244 MiB llama_params_fit_impl: context size set by user to 192000 -> no change llama_params_fit: failed to fit params to free device memory: n_gpu_layers already set by user to 99, abort llama_params_fit: fitting params to free memory took 0.97 seconds llama_model_load_from_file_impl: using device CUDA0 (NVIDIA GeForce RTX 4060 Laptop GPU) (0000:01:00.0) - 7106 MiB free llama_model_loader: loaded meta data with 66 key-value pairs and 733 tensors from models\Qwen3.5-35B-A3B-heretic-Opus-4.6-Distilled.i1-Q4_K_M.gguf (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: general.architecture str = qwen35moe llama_model_loader: - kv 1: general.type str = model llama_model_loader: - kv 2: general.name str = Qwen3.5 35B A3B Heretic Opus 4.6 Dist... llama_model_loader: - kv 3: general.version str = 4.6 llama_model_loader: - kv 4: general.finetune str = heretic-Opus-Distilled llama_model_loader: - kv 5: general.basename str = Qwen3.5 llama_model_loader: - kv 6: general.size_label str = 35B-A3B llama_model_loader: - kv 7: general.license str = apache-2.0 llama_model_loader: - kv 8: general.base_model.count u32 = 1 llama_model_loader: - kv 9: general.base_model.0.name str = Qwen3.5 35B A3B Heretic llama_model_loader: - kv 10: general.base_model.0.organization str = Jongsim llama_model_loader: - kv 11: general.base_model.0.repo_url str = https://huggingface.co/Jongsim/Qwen3.... llama_model_loader: - kv 12: general.dataset.count u32 = 3 llama_model_loader: - kv 13: general.dataset.0.name str = Opus 4.6 Reasoning 3000x Filtered llama_model_loader: - kv 14: general.dataset.0.organization str = Nohurry llama_model_loader: - kv 15: general.dataset.0.repo_url str = https://huggingface.co/nohurry/Opus-4... llama_model_loader: - kv 16: general.dataset.1.name str = Claude 4.5 Opus High Reasoning 250x llama_model_loader: - kv 17: general.dataset.1.organization str = TeichAI llama_model_loader: - kv 18: general.dataset.1.repo_url str = https://huggingface.co/TeichAI/claude... llama_model_loader: - kv 19: general.dataset.2.name str = Qwen3.5 Reasoning 700x llama_model_loader: - kv 20: general.dataset.2.organization str = Jackrong llama_model_loader: - kv 21: general.dataset.2.repo_url str = https://huggingface.co/Jackrong/Qwen3... llama_model_loader: - kv 22: general.tags arr[str,10] = ["qwen3_5_moe", "heretic", "abliterat... llama_model_loader: - kv 23: general.languages arr[str,1] = ["en"] llama_model_loader: - kv 24: qwen35moe.block_count u32 = 40 llama_model_loader: - kv 25: qwen35moe.context_length u32 = 262144 llama_model_loader: - kv 26: qwen35moe.embedding_length u32 = 2048 llama_model_loader: - kv 27: qwen35moe.attention.head_count u32 = 16 llama_model_loader: - kv 28: qwen35moe.attention.head_count_kv u32 = 2 llama_model_loader: - kv 29: qwen35moe.rope.dimension_sections arr[i32,4] = [11, 11, 10, 0] llama_model_loader: - kv 30: qwen35moe.rope.freq_base f32 = 10000000.000000 llama_model_loader: - kv 31: qwen35moe.attention.layer_norm_rms_epsilon f32 = 0.000001 llama_model_loader: - kv 32: qwen35moe.expert_count u32 = 256 llama_model_loader: - kv 33: qwen35moe.expert_used_count u32 = 8 llama_model_loader: - kv 34: qwen35moe.attention.key_length u32 = 256 llama_model_loader: - kv 35: qwen35moe.attention.value_length u32 = 256 llama_model_loader: - kv 36: qwen35moe.expert_feed_forward_length u32 = 512 llama_model_loader: - kv 37: qwen35moe.expert_shared_feed_forward_length u32 = 512 llama_model_loader: - kv 38: qwen35moe.ssm.conv_kernel u32 = 4 llama_model_loader: - kv 39: qwen35moe.ssm.state_size u32 = 128 llama_model_loader: - kv 40: qwen35moe.ssm.group_count u32 = 16 llama_model_loader: - kv 41: qwen35moe.ssm.time_step_rank u32 = 32 llama_model_loader: - kv 42: qwen35moe.ssm.inner_size u32 = 4096 llama_model_loader: - kv 43: qwen35moe.full_attention_interval u32 = 4 llama_model_loader: - kv 44: qwen35moe.rope.dimension_count u32 = 64 llama_model_loader: - kv 45: tokenizer.ggml.model str = gpt2 llama_model_loader: - kv 46: tokenizer.ggml.pre str = qwen35 llama_model_loader: - kv 47: tokenizer.ggml.tokens arr[str,248320] = ["!", "\"", "#", "$", "%", "&", "'", ... llama_model_loader: - kv 48: tokenizer.ggml.token_type arr[i32,248320] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... llama_model_loader: - kv 49: tokenizer.ggml.merges arr[str,247587] = ["─á ─á", "─á─á ─á─á", "i n", "─á t",... llama_model_loader: - kv 50: tokenizer.ggml.eos_token_id u32 = 248046 llama_model_loader: - kv 51: tokenizer.ggml.padding_token_id u32 = 248044 llama_model_loader: - kv 52: tokenizer.chat_template str = {%- set image_count = namespace(value... llama_model_loader: - kv 53: general.quantization_version u32 = 2 llama_model_loader: - kv 54: general.file_type u32 = 15 llama_model_loader: - kv 55: general.url str = https://huggingface.co/mradermacher/Q... llama_model_loader: - kv 56: mradermacher.quantize_version str = 2 llama_model_loader: - kv 57: mradermacher.quantized_by str = mradermacher llama_model_loader: - kv 58: mradermacher.quantized_at str = 2026-03-16T15:28:00+01:00 llama_model_loader: - kv 59: mradermacher.quantized_on str = nico1 llama_model_loader: - kv 60: general.source.url str = https://huggingface.co/Jongsim/Qwen3.... llama_model_loader: - kv 61: mradermacher.convert_type str = hf llama_model_loader: - kv 62: quantize.imatrix.file str = Qwen3.5-35B-A3B-heretic-Opus-4.6-Dist... llama_model_loader: - kv 63: quantize.imatrix.dataset str = imatrix-training-full-3 llama_model_loader: - kv 64: quantize.imatrix.entries_count u32 = 510 llama_model_loader: - kv 65: quantize.imatrix.chunks_count u32 = 319 llama_model_loader: - type f32: 301 tensors llama_model_loader: - type q4_K: 371 tensors llama_model_loader: - type q6_K: 61 tensors print_info: file format = GGUF V3 (latest) print_info: file type = Q4_K - Medium print_info: file size = 19.70 GiB (4.88 BPW) load: 0 unused tokens load: printing all EOG tokens: load: - 248044 ('<|endoftext|>') load: - 248046 ('<|im_end|>') load: - 248063 ('<|fim_pad|>') load: - 248064 ('<|repo_name|>') load: - 248065 ('<|file_sep|>') load: special tokens cache size = 33 load: token to piece cache size = 1.7581 MB print_info: arch = qwen35moe print_info: vocab_only = 0 print_info: no_alloc = 0 print_info: n_ctx_train = 262144 print_info: n_embd = 2048 print_info: n_embd_inp = 2048 print_info: n_layer = 40 print_info: n_head = 16 print_info: n_head_kv = 2 print_info: n_rot = 64 print_info: n_swa = 0 print_info: is_swa_any = 0 print_info: n_embd_head_k = 256 print_info: n_embd_head_v = 256 print_info: n_gqa = 8 print_info: n_embd_k_gqa = 512 print_info: n_embd_v_gqa = 512 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 = 0.0e+00 print_info: n_ff = 0 print_info: n_expert = 256 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 = 0 print_info: rope type = 40 print_info: rope scaling = linear print_info: freq_base_train = 10000000.0 print_info: freq_scale_train = 1 print_info: n_ctx_orig_yarn = 262144 print_info: rope_yarn_log_mul = 0.0000 print_info: rope_finetuned = unknown print_info: mrope sections = [11, 11, 10, 0] print_info: ssm_d_conv = 4 print_info: ssm_d_inner = 4096 print_info: ssm_d_state = 128 print_info: ssm_dt_rank = 32 print_info: ssm_n_group = 16 print_info: ssm_dt_b_c_rms = 0 print_info: model type = 35B.A3B print_info: model params = 34.66 B print_info: general.name = Qwen3.5 35B A3B Heretic Opus 4.6 Distilled print_info: vocab type = BPE print_info: n_vocab = 248320 print_info: n_merges = 247587 print_info: BOS token = 11 ',' print_info: EOS token = 248046 '<|im_end|>' print_info: EOT token = 248046 '<|im_end|>' print_info: PAD token = 248044 '<|endoftext|>' print_info: LF token = 198 '─è' print_info: FIM PRE token = 248060 '<|fim_prefix|>' print_info: FIM SUF token = 248062 '<|fim_suffix|>' print_info: FIM MID token = 248061 '<|fim_middle|>' print_info: FIM PAD token = 248063 '<|fim_pad|>' print_info: FIM REP token = 248064 '<|repo_name|>' print_info: FIM SEP token = 248065 '<|file_sep|>' print_info: EOG token = 248044 '<|endoftext|>' print_info: EOG token = 248046 '<|im_end|>' print_info: EOG token = 248063 '<|fim_pad|>' print_info: EOG token = 248064 '<|repo_name|>' print_info: EOG token = 248065 '<|file_sep|>' print_info: max token length = 256 load_tensors: loading model tensors, this can take a while... (mmap = true, direct_io = false) load_tensors: offloading output layer to GPU load_tensors: offloading 39 repeating layers to GPU load_tensors: offloaded 41/41 layers to GPU load_tensors: CPU_Mapped model buffer size = 19777.28 MiB load_tensors: CUDA0 model buffer size = 1302.90 MiB ................................................................................................ common_init_result: added <|endoftext|> logit bias = -inf common_init_result: added <|im_end|> logit bias = -inf common_init_result: added <|fim_pad|> logit bias = -inf common_init_result: added <|repo_name|> logit bias = -inf common_init_result: added <|file_sep|> logit bias = -inf llama_context: constructing llama_context llama_context: n_seq_max = 4 llama_context: n_ctx = 192000 llama_context: n_ctx_seq = 192000 llama_context: n_batch = 4096 llama_context: n_ubatch = 2048 llama_context: causal_attn = 1 llama_context: flash_attn = enabled llama_context: kv_unified = true llama_context: freq_base = 10000000.0 llama_context: freq_scale = 1 llama_context: n_ctx_seq (192000) < n_ctx_train (262144) -- the full capacity of the model will not be utilized llama_context: CUDA_Host output buffer size = 3.79 MiB llama_kv_cache: CUDA0 KV buffer size = 1992.19 MiB llama_kv_cache: size = 1992.19 MiB (192000 cells, 10 layers, 4/1 seqs), K (q8_0): 996.09 MiB, V (q8_0): 996.09 MiB llama_memory_recurrent: CUDA0 RS buffer size = 251.25 MiB llama_memory_recurrent: size = 251.25 MiB ( 4 cells, 40 layers, 4 seqs), R (f32): 11.25 MiB, S (f32): 240.00 MiB sched_reserve: reserving ... 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: CUDA0 compute buffer size = 2528.00 MiB sched_reserve: CUDA_Host compute buffer size = 1532.08 MiB sched_reserve: graph nodes = 3729 sched_reserve: graph splits = 122 (with bs=2048), 82 (with bs=1) sched_reserve: reserve took 235.90 ms, sched copies = 1 common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable) clip_model_loader: model name: Qwen3.5-35B-A3B clip_model_loader: description: clip_model_loader: GGUF version: 3 clip_model_loader: alignment: 32 clip_model_loader: n_tensors: 334 clip_model_loader: n_kv: 35 clip_model_loader: has vision encoder clip_ctx: CLIP using CUDA0 backend load_hparams: Qwen-VL models require at minimum 1024 image tokens to function correctly on grounding tasks load_hparams: if you encounter problems with accuracy, try adding --image-min-tokens 1024 load_hparams: more info: https://github.com/ggml-org/llama.cpp/issues/16842 load_hparams: projector: qwen3vl_merger 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 load_hparams: projection_dim: 2048 --- vision hparams --- load_hparams: image_size: 768 load_hparams: patch_size: 16 load_hparams: has_llava_proj: 0 load_hparams: minicpmv_version: 0 load_hparams: n_merge: 2 load_hparams: n_wa_pattern: 0 load_hparams: image_min_pixels: 8192 load_hparams: image_max_pixels: 4194304 load_hparams: model size: 1703.53 MiB load_hparams: metadata size: 0.12 MiB warmup: warmup with image size = 1472 x 1472 alloc_compute_meta: CUDA0 compute buffer size = 248.10 MiB alloc_compute_meta: CPU compute buffer size = 24.93 MiB alloc_compute_meta: graph splits = 1, nodes = 823 warmup: flash attention is enabled srv load_model: loaded multimodal model, 'models\mmproj-F32.gguf' srv load_model: initializing slots, n_slots = 4 common_speculative_is_compat: the target context does not support partial sequence removal srv load_model: speculative decoding not supported by this context slot load_model: id 0 | task -1 | new slot, n_ctx = 192000 slot load_model: id 1 | task -1 | new slot, n_ctx = 192000 slot load_model: id 2 | task -1 | new slot, n_ctx = 192000 slot load_model: id 3 | task -1 | new slot, n_ctx = 192000 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 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 = 1 main: model loaded main: server is listening on http://127.0.0.1:8080 main: starting the main loop... srv update_slots: all slots are idle