github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-15 12:48:05 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #12757] Qwen3-Embedding-8B "model does not support embeddings" #8461

New Issue
Closed
opened 2026-04-12 21:09:04 -05:00 by GiteaMirror · 26 comments
GiteaMirror commented 2026-04-12 21:09:04 -05:00
Owner
Copy Link

Originally created by @ndrewpj on GitHub (Oct 23, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/12757

Originally assigned to: @npardal on GitHub.

What is the issue?

Hi!
After updating Ollama tp latest v.0.12.6 suddenly I got my Qwen3 model not able to make embeddings. The Ollama logs also show that.
The model is hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q6_K
CUDA 12.9
4 x gtx 1080
Ollama config:
sudo docker run -d --gpus=all --network=bridge -v ollama:/root/.ollama -p 27171:27171 --name ollama2 -e OLLAMA_HOST=0.0.0.0:27171 -e OLLAMA_MAX_QUEUE=512 -e OLLAMA_MAX_PARALLEL=4 -e OLLAMA_NUM_PARALLEL=2 -e OLLAMA_FLASH_ATTENTION=1 -e OLLAMA_SCHED_SPREAD=1 -e OLLAMA_MAX_LOADED_MODELS=12 ollama/ollama:latest

Image

Relevant log output

time=2025-10-23T17:21:20.521Z level=INFO source=routes.go:1511 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: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:true OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://0.0.0.0:27171 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:12 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:/root/.ollama/models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:2 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:true ROCR_VISIBLE_DEVICES: http_proxy: https_proxy: no_proxy:]"

time=2025-10-23T17:21:20.526Z level=INFO source=images.go:522 msg="total blobs: 70"

time=2025-10-23T17:21:20.529Z level=INFO source=images.go:529 msg="total unused blobs removed: 0"

time=2025-10-23T17:21:20.530Z level=INFO source=routes.go:1564 msg="Listening on [::]:27171 (version 0.12.6)"

time=2025-10-23T17:21:20.530Z level=INFO source=runner.go:80 msg="discovering available GPUs..."

time=2025-10-23T17:21:21.575Z level=INFO source=types.go:112 msg="inference compute" id=GPU-56195da2-4f22-31d4-3d54-2558308cce33 library=CUDA compute=6.1 name=CUDA0 description="NVIDIA GeForce GTX 1080" libdirs=ollama,cuda_v12 driver=13.0 pci_id=17:00.0 type=discrete total="8.0 GiB" available="7.0 GiB"

time=2025-10-23T17:21:21.575Z level=INFO source=types.go:112 msg="inference compute" id=GPU-f5656f9d-f750-011c-bafa-b3d184dc11f1 library=CUDA compute=6.1 name=CUDA1 description="NVIDIA GeForce GTX 1080" libdirs=ollama,cuda_v12 driver=13.0 pci_id=18:00.0 type=discrete total="8.0 GiB" available="7.8 GiB"

time=2025-10-23T17:21:21.575Z level=INFO source=types.go:112 msg="inference compute" id=GPU-ad59abf3-adcd-44c1-0376-4c6778a0c4bb library=CUDA compute=6.1 name=CUDA2 description="NVIDIA GeForce GTX 1080" libdirs=ollama,cuda_v12 driver=13.0 pci_id=65:00.0 type=discrete total="8.0 GiB" available="7.8 GiB"

time=2025-10-23T17:21:21.575Z level=INFO source=types.go:112 msg="inference compute" id=GPU-67108390-cf1c-53c0-6eac-58a2f1a86871 library=CUDA compute=6.1 name=CUDA3 description="NVIDIA GeForce GTX 1080" libdirs=ollama,cuda_v12 driver=13.0 pci_id=b3:00.0 type=discrete total="8.0 GiB" available="7.8 GiB"

time=2025-10-23T17:22:16.307Z level=INFO source=server.go:216 msg="enabling flash attention"

time=2025-10-23T17:22:16.307Z level=INFO source=server.go:400 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --model /root/.ollama/models/blobs/sha256-2578fb0291be2133804d1ebff2840feddfbc53d662c9031fbe33931710911ce9 --port 40649"

time=2025-10-23T17:22:16.308Z level=INFO source=server.go:676 msg="loading model" "model layers"=37 requested=-1

time=2025-10-23T17:22:16.309Z level=INFO source=server.go:682 msg="system memory" total="62.5 GiB" free="45.8 GiB" free_swap="38.0 GiB"

time=2025-10-23T17:22:16.309Z level=INFO source=server.go:690 msg="gpu memory" id=GPU-56195da2-4f22-31d4-3d54-2558308cce33 library=CUDA available="6.6 GiB" free="7.0 GiB" minimum="457.0 MiB" overhead="0 B"

time=2025-10-23T17:22:16.309Z level=INFO source=server.go:690 msg="gpu memory" id=GPU-f5656f9d-f750-011c-bafa-b3d184dc11f1 library=CUDA available="7.4 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B"

time=2025-10-23T17:22:16.309Z level=INFO source=server.go:690 msg="gpu memory" id=GPU-ad59abf3-adcd-44c1-0376-4c6778a0c4bb library=CUDA available="7.4 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B"

time=2025-10-23T17:22:16.309Z level=INFO source=server.go:690 msg="gpu memory" id=GPU-67108390-cf1c-53c0-6eac-58a2f1a86871 library=CUDA available="7.4 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B"

time=2025-10-23T17:22:16.322Z level=INFO source=runner.go:1332 msg="starting ollama engine"

time=2025-10-23T17:22:16.322Z level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:40649"

time=2025-10-23T17:22:16.331Z level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:2 BatchSize:512 FlashAttention:true KvSize:16384 KvCacheType: NumThreads:10 GPULayers:37[ID:GPU-56195da2-4f22-31d4-3d54-2558308cce33 Layers:37(0..36)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"

time=2025-10-23T17:22:16.360Z level=INFO source=ggml.go:134 msg="" architecture=qwen3 file_type=Q6_K name="Qwen3 Embedding 8B Bf16" description="" num_tensors=398 num_key_values=28

load_backend: loaded CPU backend from /usr/lib/ollama/libggml-cpu-skylakex.so

ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no

ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no

ggml_cuda_init: found 4 CUDA devices:

  Device 0: NVIDIA GeForce GTX 1080, compute capability 6.1, VMM: yes, ID: GPU-56195da2-4f22-31d4-3d54-2558308cce33

  Device 1: NVIDIA GeForce GTX 1080, compute capability 6.1, VMM: yes, ID: GPU-f5656f9d-f750-011c-bafa-b3d184dc11f1

  Device 2: NVIDIA GeForce GTX 1080, compute capability 6.1, VMM: yes, ID: GPU-ad59abf3-adcd-44c1-0376-4c6778a0c4bb

  Device 3: NVIDIA GeForce GTX 1080, compute capability 6.1, VMM: yes, ID: GPU-67108390-cf1c-53c0-6eac-58a2f1a86871

load_backend: loaded CUDA backend from /usr/lib/ollama/cuda_v12/libggml-cuda.so

time=2025-10-23T17:22:16.435Z level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=500,520,600,610,700,750,800,860,870,890,900,1200 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 CUDA.1.ARCHS=500,520,600,610,700,750,800,860,870,890,900,1200 CUDA.1.USE_GRAPHS=1 CUDA.1.PEER_MAX_BATCH_SIZE=128 CUDA.2.ARCHS=500,520,600,610,700,750,800,860,870,890,900,1200 CUDA.2.USE_GRAPHS=1 CUDA.2.PEER_MAX_BATCH_SIZE=128 CUDA.3.ARCHS=500,520,600,610,700,750,800,860,870,890,900,1200 CUDA.3.USE_GRAPHS=1 CUDA.3.PEER_MAX_BATCH_SIZE=128 compiler=cgo(gcc)

time=2025-10-23T17:22:16.556Z level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:2 BatchSize:512 FlashAttention:true KvSize:16384 KvCacheType: NumThreads:10 GPULayers:37[ID:GPU-f5656f9d-f750-011c-bafa-b3d184dc11f1 Layers:10(0..9) ID:GPU-67108390-cf1c-53c0-6eac-58a2f1a86871 Layers:10(10..19) ID:GPU-ad59abf3-adcd-44c1-0376-4c6778a0c4bb Layers:10(20..29) ID:GPU-56195da2-4f22-31d4-3d54-2558308cce33 Layers:7(30..36)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"

time=2025-10-23T17:22:16.867Z level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:2 BatchSize:512 FlashAttention:true KvSize:16384 KvCacheType: NumThreads:10 GPULayers:37[ID:GPU-f5656f9d-f750-011c-bafa-b3d184dc11f1 Layers:10(0..9) ID:GPU-67108390-cf1c-53c0-6eac-58a2f1a86871 Layers:10(10..19) ID:GPU-ad59abf3-adcd-44c1-0376-4c6778a0c4bb Layers:10(20..29) ID:GPU-56195da2-4f22-31d4-3d54-2558308cce33 Layers:7(30..36)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"

time=2025-10-23T17:22:16.927Z level=INFO source=runner.go:1205 msg=load request="{Operation:commit LoraPath:[] Parallel:2 BatchSize:512 FlashAttention:true KvSize:16384 KvCacheType: NumThreads:10 GPULayers:37[ID:GPU-f5656f9d-f750-011c-bafa-b3d184dc11f1 Layers:10(0..9) ID:GPU-67108390-cf1c-53c0-6eac-58a2f1a86871 Layers:10(10..19) ID:GPU-ad59abf3-adcd-44c1-0376-4c6778a0c4bb Layers:10(20..29) ID:GPU-56195da2-4f22-31d4-3d54-2558308cce33 Layers:7(30..36)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"

time=2025-10-23T17:22:16.927Z level=INFO source=ggml.go:480 msg="offloading 36 repeating layers to GPU"

time=2025-10-23T17:22:16.927Z level=INFO source=ggml.go:487 msg="offloading output layer to GPU"

time=2025-10-23T17:22:16.927Z level=INFO source=ggml.go:492 msg="offloaded 37/37 layers to GPU"

time=2025-10-23T17:22:16.927Z level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="1.4 GiB"

time=2025-10-23T17:22:16.928Z level=INFO source=device.go:206 msg="model weights" device=CUDA1 size="1.5 GiB"

time=2025-10-23T17:22:16.928Z level=INFO source=device.go:206 msg="model weights" device=CUDA2 size="1.5 GiB"

time=2025-10-23T17:22:16.928Z level=INFO source=device.go:206 msg="model weights" device=CUDA3 size="1.5 GiB"

time=2025-10-23T17:22:16.928Z level=INFO source=device.go:211 msg="model weights" device=CPU size="486.0 MiB"

time=2025-10-23T17:22:16.928Z level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="384.0 MiB"

time=2025-10-23T17:22:16.928Z level=INFO source=device.go:217 msg="kv cache" device=CUDA1 size="640.0 MiB"

time=2025-10-23T17:22:16.928Z level=INFO source=device.go:217 msg="kv cache" device=CUDA2 size="640.0 MiB"

time=2025-10-23T17:22:16.928Z level=INFO source=device.go:217 msg="kv cache" device=CUDA3 size="640.0 MiB"

time=2025-10-23T17:22:16.928Z level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="122.0 MiB"

time=2025-10-23T17:22:16.928Z level=INFO source=device.go:228 msg="compute graph" device=CUDA1 size="130.0 MiB"

time=2025-10-23T17:22:16.928Z level=INFO source=device.go:228 msg="compute graph" device=CUDA2 size="122.0 MiB"

time=2025-10-23T17:22:16.928Z level=INFO source=device.go:228 msg="compute graph" device=CUDA3 size="122.0 MiB"

time=2025-10-23T17:22:16.928Z level=INFO source=device.go:233 msg="compute graph" device=CPU size="8.0 MiB"

time=2025-10-23T17:22:16.928Z level=INFO source=device.go:238 msg="total memory" size="9.0 GiB"

time=2025-10-23T17:22:16.928Z level=INFO source=sched.go:482 msg="loaded runners" count=1

time=2025-10-23T17:22:16.928Z level=INFO source=server.go:1272 msg="waiting for llama runner to start responding"

time=2025-10-23T17:22:16.929Z level=INFO source=server.go:1306 msg="waiting for server to become available" status="llm server loading model"

time=2025-10-23T17:22:17.933Z level=INFO source=server.go:1310 msg="llama runner started in 1.63 seconds"

[GIN] 2025/10/23 - 17:22:17 | 500 |  2.373599419s |      172.17.0.1 | POST     "/api/embed"

time=2025-10-23T17:22:17.988Z level=INFO source=server.go:1635 msg="llm embedding error: this model does not support embeddings"

time=2025-10-23T17:22:18.097Z level=INFO source=server.go:1635 msg="llm embedding error: this model does not support embeddings"

[GIN] 2025/10/23 - 17:22:18 | 500 |   95.388541ms |      172.17.0.1 | POST     "/api/embed"

OS

Linux

GPU

Nvidia

CPU

Intel

Ollama version

0.12.6

Originally created by @ndrewpj on GitHub (Oct 23, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/12757 Originally assigned to: @npardal on GitHub. ### What is the issue? Hi! After updating Ollama tp latest v.0.12.6 suddenly I got my Qwen3 model not able to make embeddings. The Ollama logs also show that. The model is hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q6_K CUDA 12.9 4 x gtx 1080 Ollama config: sudo docker run -d --gpus=all --network=bridge -v ollama:/root/.ollama -p 27171:27171 --name ollama2 -e OLLAMA_HOST=0.0.0.0:27171 -e OLLAMA_MAX_QUEUE=512 -e OLLAMA_MAX_PARALLEL=4 -e OLLAMA_NUM_PARALLEL=2 -e OLLAMA_FLASH_ATTENTION=1 -e OLLAMA_SCHED_SPREAD=1 -e OLLAMA_MAX_LOADED_MODELS=12 ollama/ollama:latest <img width="808" height="845" alt="Image" src="https://github.com/user-attachments/assets/7859b645-f78a-4e9a-8519-5d7a3f5a530b" /> ### Relevant log output ```shell time=2025-10-23T17:21:20.521Z level=INFO source=routes.go:1511 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: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:true OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://0.0.0.0:27171 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:12 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:/root/.ollama/models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:2 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:true ROCR_VISIBLE_DEVICES: http_proxy: https_proxy: no_proxy:]" time=2025-10-23T17:21:20.526Z level=INFO source=images.go:522 msg="total blobs: 70" time=2025-10-23T17:21:20.529Z level=INFO source=images.go:529 msg="total unused blobs removed: 0" time=2025-10-23T17:21:20.530Z level=INFO source=routes.go:1564 msg="Listening on [::]:27171 (version 0.12.6)" time=2025-10-23T17:21:20.530Z level=INFO source=runner.go:80 msg="discovering available GPUs..." time=2025-10-23T17:21:21.575Z level=INFO source=types.go:112 msg="inference compute" id=GPU-56195da2-4f22-31d4-3d54-2558308cce33 library=CUDA compute=6.1 name=CUDA0 description="NVIDIA GeForce GTX 1080" libdirs=ollama,cuda_v12 driver=13.0 pci_id=17:00.0 type=discrete total="8.0 GiB" available="7.0 GiB" time=2025-10-23T17:21:21.575Z level=INFO source=types.go:112 msg="inference compute" id=GPU-f5656f9d-f750-011c-bafa-b3d184dc11f1 library=CUDA compute=6.1 name=CUDA1 description="NVIDIA GeForce GTX 1080" libdirs=ollama,cuda_v12 driver=13.0 pci_id=18:00.0 type=discrete total="8.0 GiB" available="7.8 GiB" time=2025-10-23T17:21:21.575Z level=INFO source=types.go:112 msg="inference compute" id=GPU-ad59abf3-adcd-44c1-0376-4c6778a0c4bb library=CUDA compute=6.1 name=CUDA2 description="NVIDIA GeForce GTX 1080" libdirs=ollama,cuda_v12 driver=13.0 pci_id=65:00.0 type=discrete total="8.0 GiB" available="7.8 GiB" time=2025-10-23T17:21:21.575Z level=INFO source=types.go:112 msg="inference compute" id=GPU-67108390-cf1c-53c0-6eac-58a2f1a86871 library=CUDA compute=6.1 name=CUDA3 description="NVIDIA GeForce GTX 1080" libdirs=ollama,cuda_v12 driver=13.0 pci_id=b3:00.0 type=discrete total="8.0 GiB" available="7.8 GiB" time=2025-10-23T17:22:16.307Z level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-23T17:22:16.307Z level=INFO source=server.go:400 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --model /root/.ollama/models/blobs/sha256-2578fb0291be2133804d1ebff2840feddfbc53d662c9031fbe33931710911ce9 --port 40649" time=2025-10-23T17:22:16.308Z level=INFO source=server.go:676 msg="loading model" "model layers"=37 requested=-1 time=2025-10-23T17:22:16.309Z level=INFO source=server.go:682 msg="system memory" total="62.5 GiB" free="45.8 GiB" free_swap="38.0 GiB" time=2025-10-23T17:22:16.309Z level=INFO source=server.go:690 msg="gpu memory" id=GPU-56195da2-4f22-31d4-3d54-2558308cce33 library=CUDA available="6.6 GiB" free="7.0 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-23T17:22:16.309Z level=INFO source=server.go:690 msg="gpu memory" id=GPU-f5656f9d-f750-011c-bafa-b3d184dc11f1 library=CUDA available="7.4 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-23T17:22:16.309Z level=INFO source=server.go:690 msg="gpu memory" id=GPU-ad59abf3-adcd-44c1-0376-4c6778a0c4bb library=CUDA available="7.4 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-23T17:22:16.309Z level=INFO source=server.go:690 msg="gpu memory" id=GPU-67108390-cf1c-53c0-6eac-58a2f1a86871 library=CUDA available="7.4 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-23T17:22:16.322Z level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-23T17:22:16.322Z level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:40649" time=2025-10-23T17:22:16.331Z level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:2 BatchSize:512 FlashAttention:true KvSize:16384 KvCacheType: NumThreads:10 GPULayers:37[ID:GPU-56195da2-4f22-31d4-3d54-2558308cce33 Layers:37(0..36)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-23T17:22:16.360Z level=INFO source=ggml.go:134 msg="" architecture=qwen3 file_type=Q6_K name="Qwen3 Embedding 8B Bf16" description="" num_tensors=398 num_key_values=28 load_backend: loaded CPU backend from /usr/lib/ollama/libggml-cpu-skylakex.so ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 4 CUDA devices: Device 0: NVIDIA GeForce GTX 1080, compute capability 6.1, VMM: yes, ID: GPU-56195da2-4f22-31d4-3d54-2558308cce33 Device 1: NVIDIA GeForce GTX 1080, compute capability 6.1, VMM: yes, ID: GPU-f5656f9d-f750-011c-bafa-b3d184dc11f1 Device 2: NVIDIA GeForce GTX 1080, compute capability 6.1, VMM: yes, ID: GPU-ad59abf3-adcd-44c1-0376-4c6778a0c4bb Device 3: NVIDIA GeForce GTX 1080, compute capability 6.1, VMM: yes, ID: GPU-67108390-cf1c-53c0-6eac-58a2f1a86871 load_backend: loaded CUDA backend from /usr/lib/ollama/cuda_v12/libggml-cuda.so time=2025-10-23T17:22:16.435Z level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=500,520,600,610,700,750,800,860,870,890,900,1200 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 CUDA.1.ARCHS=500,520,600,610,700,750,800,860,870,890,900,1200 CUDA.1.USE_GRAPHS=1 CUDA.1.PEER_MAX_BATCH_SIZE=128 CUDA.2.ARCHS=500,520,600,610,700,750,800,860,870,890,900,1200 CUDA.2.USE_GRAPHS=1 CUDA.2.PEER_MAX_BATCH_SIZE=128 CUDA.3.ARCHS=500,520,600,610,700,750,800,860,870,890,900,1200 CUDA.3.USE_GRAPHS=1 CUDA.3.PEER_MAX_BATCH_SIZE=128 compiler=cgo(gcc) time=2025-10-23T17:22:16.556Z level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:2 BatchSize:512 FlashAttention:true KvSize:16384 KvCacheType: NumThreads:10 GPULayers:37[ID:GPU-f5656f9d-f750-011c-bafa-b3d184dc11f1 Layers:10(0..9) ID:GPU-67108390-cf1c-53c0-6eac-58a2f1a86871 Layers:10(10..19) ID:GPU-ad59abf3-adcd-44c1-0376-4c6778a0c4bb Layers:10(20..29) ID:GPU-56195da2-4f22-31d4-3d54-2558308cce33 Layers:7(30..36)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-23T17:22:16.867Z level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:2 BatchSize:512 FlashAttention:true KvSize:16384 KvCacheType: NumThreads:10 GPULayers:37[ID:GPU-f5656f9d-f750-011c-bafa-b3d184dc11f1 Layers:10(0..9) ID:GPU-67108390-cf1c-53c0-6eac-58a2f1a86871 Layers:10(10..19) ID:GPU-ad59abf3-adcd-44c1-0376-4c6778a0c4bb Layers:10(20..29) ID:GPU-56195da2-4f22-31d4-3d54-2558308cce33 Layers:7(30..36)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-23T17:22:16.927Z level=INFO source=runner.go:1205 msg=load request="{Operation:commit LoraPath:[] Parallel:2 BatchSize:512 FlashAttention:true KvSize:16384 KvCacheType: NumThreads:10 GPULayers:37[ID:GPU-f5656f9d-f750-011c-bafa-b3d184dc11f1 Layers:10(0..9) ID:GPU-67108390-cf1c-53c0-6eac-58a2f1a86871 Layers:10(10..19) ID:GPU-ad59abf3-adcd-44c1-0376-4c6778a0c4bb Layers:10(20..29) ID:GPU-56195da2-4f22-31d4-3d54-2558308cce33 Layers:7(30..36)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-23T17:22:16.927Z level=INFO source=ggml.go:480 msg="offloading 36 repeating layers to GPU" time=2025-10-23T17:22:16.927Z level=INFO source=ggml.go:487 msg="offloading output layer to GPU" time=2025-10-23T17:22:16.927Z level=INFO source=ggml.go:492 msg="offloaded 37/37 layers to GPU" time=2025-10-23T17:22:16.927Z level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="1.4 GiB" time=2025-10-23T17:22:16.928Z level=INFO source=device.go:206 msg="model weights" device=CUDA1 size="1.5 GiB" time=2025-10-23T17:22:16.928Z level=INFO source=device.go:206 msg="model weights" device=CUDA2 size="1.5 GiB" time=2025-10-23T17:22:16.928Z level=INFO source=device.go:206 msg="model weights" device=CUDA3 size="1.5 GiB" time=2025-10-23T17:22:16.928Z level=INFO source=device.go:211 msg="model weights" device=CPU size="486.0 MiB" time=2025-10-23T17:22:16.928Z level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="384.0 MiB" time=2025-10-23T17:22:16.928Z level=INFO source=device.go:217 msg="kv cache" device=CUDA1 size="640.0 MiB" time=2025-10-23T17:22:16.928Z level=INFO source=device.go:217 msg="kv cache" device=CUDA2 size="640.0 MiB" time=2025-10-23T17:22:16.928Z level=INFO source=device.go:217 msg="kv cache" device=CUDA3 size="640.0 MiB" time=2025-10-23T17:22:16.928Z level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="122.0 MiB" time=2025-10-23T17:22:16.928Z level=INFO source=device.go:228 msg="compute graph" device=CUDA1 size="130.0 MiB" time=2025-10-23T17:22:16.928Z level=INFO source=device.go:228 msg="compute graph" device=CUDA2 size="122.0 MiB" time=2025-10-23T17:22:16.928Z level=INFO source=device.go:228 msg="compute graph" device=CUDA3 size="122.0 MiB" time=2025-10-23T17:22:16.928Z level=INFO source=device.go:233 msg="compute graph" device=CPU size="8.0 MiB" time=2025-10-23T17:22:16.928Z level=INFO source=device.go:238 msg="total memory" size="9.0 GiB" time=2025-10-23T17:22:16.928Z level=INFO source=sched.go:482 msg="loaded runners" count=1 time=2025-10-23T17:22:16.928Z level=INFO source=server.go:1272 msg="waiting for llama runner to start responding" time=2025-10-23T17:22:16.929Z level=INFO source=server.go:1306 msg="waiting for server to become available" status="llm server loading model" time=2025-10-23T17:22:17.933Z level=INFO source=server.go:1310 msg="llama runner started in 1.63 seconds" [GIN] 2025/10/23 - 17:22:17 | 500 | 2.373599419s | 172.17.0.1 | POST "/api/embed" time=2025-10-23T17:22:17.988Z level=INFO source=server.go:1635 msg="llm embedding error: this model does not support embeddings" time=2025-10-23T17:22:18.097Z level=INFO source=server.go:1635 msg="llm embedding error: this model does not support embeddings" [GIN] 2025/10/23 - 17:22:18 | 500 | 95.388541ms | 172.17.0.1 | POST "/api/embed" ``` ### OS Linux ### GPU Nvidia ### CPU Intel ### Ollama version 0.12.6
GiteaMirror added the needs more infobug labels 2026-04-12 21:09:04 -05:00
GiteaMirror commented 2026-04-12 21:09:06 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Oct 23, 2025):

I wasn't able to duplicate this.

From my testing:

% curl localhost:11434/api/embed -d '{"model": "hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q6_K", "input": "hey there"}'
{"model":"hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q6_K","embeddings":[[0.031040814,0.0064123375,-0.012913328,-0.04292166,-0.0017876483, ...

The code where the error is happening looks for the qwen3:pooling_type parameter in the config of the GGUF, and for some reason yours isn't populated. Are you certain you actually pulled the model?

<!-- gh-comment-id:3439375019 --> @pdevine commented on GitHub (Oct 23, 2025): I wasn't able to duplicate this. From my testing: ``` % curl localhost:11434/api/embed -d '{"model": "hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q6_K", "input": "hey there"}' {"model":"hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q6_K","embeddings":[[0.031040814,0.0064123375,-0.012913328,-0.04292166,-0.0017876483, ... ``` The code where the error is happening looks for the `qwen3:pooling_type` parameter in the config of the GGUF, and for some reason yours isn't populated. Are you certain you actually pulled the model?
GiteaMirror commented 2026-04-12 21:09:06 -05:00
Author
Owner
Copy Link

@ndrewpj commented on GitHub (Oct 23, 2025):

I wasn't able to duplicate this.

From my testing:

% curl localhost:11434/api/embed -d '{"model": "hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q6_K", "input": "hey there"}'
{"model":"hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q6_K","embeddings":[[0.031040814,0.0064123375,-0.012913328,-0.04292166,-0.0017876483, ...

The code where the error is happening looks for the qwen3:pooling_type parameter in the config of the GGUF, and for some reason yours isn't populated. Are you certain you actually pulled the model?

Yes, I was using this model for half a year. What should I do - re-download it or? I was trying to add some environment variables for Ollama, the new engine, ome KV cache q4_0 - could it be because of that?

<!-- gh-comment-id:3439536716 --> @ndrewpj commented on GitHub (Oct 23, 2025): > I wasn't able to duplicate this. > > From my testing: > > ``` > % curl localhost:11434/api/embed -d '{"model": "hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q6_K", "input": "hey there"}' > {"model":"hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q6_K","embeddings":[[0.031040814,0.0064123375,-0.012913328,-0.04292166,-0.0017876483, ... > ``` > > The code where the error is happening looks for the `qwen3:pooling_type` parameter in the config of the GGUF, and for some reason yours isn't populated. Are you certain you actually pulled the model? Yes, I was using this model for half a year. What should I do - re-download it or? I was trying to add some environment variables for Ollama, the new engine, ome KV cache q4_0 - could it be because of that?
GiteaMirror commented 2026-04-12 21:09:07 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Oct 23, 2025):

From this page it looks like it was updated 3 months ago?

OLLAMA_NEW_ENGINE=1 will force it to run the ollama implementation of the model, whereas OLLAMA_NEW_ENGINE=0 will force it to run on the legacy llama.cpp runner. You could try forcing it to the ollama implementation and see if that works (which I think it should, but it's not a model that we published).

<!-- gh-comment-id:3439611870 --> @pdevine commented on GitHub (Oct 23, 2025): From [this page](https://huggingface.co/Qwen/Qwen3-Embedding-8B-GGUF/tree/main) it looks like it was updated 3 months ago? `OLLAMA_NEW_ENGINE=1` will force it to run the ollama implementation of the model, whereas `OLLAMA_NEW_ENGINE=0` will force it to run on the legacy llama.cpp runner. You could try forcing it to the ollama implementation and see if that works (which I think it should, but it's not a model that we published).
GiteaMirror commented 2026-04-12 21:09:08 -05:00
Author
Owner
Copy Link

@ndrewpj commented on GitHub (Oct 24, 2025):

From this page it looks like it was updated 3 months ago?

OLLAMA_NEW_ENGINE=1 will force it to run the ollama implementation of the model, whereas OLLAMA_NEW_ENGINE=0 will force it to run on the legacy llama.cpp runner. You could try forcing it to the ollama implementation and see if that works (which I think it should, but it's not a model that we published).

3 months, let it be 3 months. Its not critical. I tried all combinations of environmental keys, no luck. Tomorrow I'll try to rollback to 0.12.5

<!-- gh-comment-id:3439937149 --> @ndrewpj commented on GitHub (Oct 24, 2025): > From [this page](https://huggingface.co/Qwen/Qwen3-Embedding-8B-GGUF/tree/main) it looks like it was updated 3 months ago? > > `OLLAMA_NEW_ENGINE=1` will force it to run the ollama implementation of the model, whereas `OLLAMA_NEW_ENGINE=0` will force it to run on the legacy llama.cpp runner. You could try forcing it to the ollama implementation and see if that works (which I think it should, but it's not a model that we published). 3 months, let it be 3 months. Its not critical. I tried all combinations of environmental keys, no luck. Tomorrow I'll try to rollback to 0.12.5
GiteaMirror commented 2026-04-12 21:09:09 -05:00
Author
Owner
Copy Link

@ndrewpj commented on GitHub (Oct 24, 2025):

So yeah, going back to v.0.12.5 fixed that. The model works now again

<!-- gh-comment-id:3441329308 --> @ndrewpj commented on GitHub (Oct 24, 2025): So yeah, going back to v.0.12.5 fixed that. The model works now again
GiteaMirror commented 2026-04-12 21:09:10 -05:00
Author
Owner
Copy Link

@naabbasi commented on GitHub (Oct 25, 2025):

I have updated the ollama to 1.12.6 and with qwen3 model, I m getting the following exception:

java.lang.RuntimeException: status code: 500; body: {"error":"this model does not support embeddings"}

Ollama is giving the 500 (Http Status)

<!-- gh-comment-id:3446725779 --> @naabbasi commented on GitHub (Oct 25, 2025): I have updated the ollama to 1.12.6 and with [qwen3](qwen3:1.7b) model, I m getting the following exception: java.lang.RuntimeException: status code: 500; body: {"error":"this model does not support embeddings"} Ollama is giving the 500 (Http Status)
GiteaMirror commented 2026-04-12 21:09:11 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Oct 25, 2025):

@naabbasi The non-embedding qwen3 model does not support embeddings. Use the qwen3-embedding model instead.

<!-- gh-comment-id:3446997713 --> @pdevine commented on GitHub (Oct 25, 2025): @naabbasi The non-embedding qwen3 model does not support embeddings. Use the `qwen3-embedding` model instead.
GiteaMirror commented 2026-04-12 21:09:11 -05:00
Author
Owner
Copy Link

@ndrewpj commented on GitHub (Oct 26, 2025):

@naabbasi The non-embedding qwen3 model does not support embeddings. Use the qwen3-embedding model instead.

Well, I got the same error with qwen3 embeddings model

<!-- gh-comment-id:3448505672 --> @ndrewpj commented on GitHub (Oct 26, 2025): > [@naabbasi](https://github.com/naabbasi) The non-embedding qwen3 model does not support embeddings. Use the `qwen3-embedding` model instead. Well, I got the same error with qwen3 embeddings model
GiteaMirror commented 2026-04-12 21:09:12 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Oct 26, 2025):

@ndrewpj is that with the ollama version of the qwen3-embeddings model or one you got somewhere else?

<!-- gh-comment-id:3448754749 --> @pdevine commented on GitHub (Oct 26, 2025): @ndrewpj is that with the ollama version of the qwen3-embeddings model or one you got somewhere else?
GiteaMirror commented 2026-04-12 21:09:12 -05:00
Author
Owner
Copy Link

@Imitater967 commented on GitHub (Oct 28, 2025):

the same issue with dengcao/Qwen3-Embedding-8B:F16

<!-- gh-comment-id:3455077088 --> @Imitater967 commented on GitHub (Oct 28, 2025): the same issue with dengcao/Qwen3-Embedding-8B:F16
GiteaMirror commented 2026-04-12 21:09:13 -05:00
Author
Owner
Copy Link

@Imitater967 commented on GitHub (Oct 28, 2025):

time=2025-10-28T16:01:25.819+08:00 level=INFO source=server.go:1272 msg="waiting for llama runner to start responding"
time=2025-10-28T16:01:25.819+08:00 level=INFO source=server.go:1306 msg="waiting for server to become available" status="llm server loading model"
time=2025-10-28T16:01:29.857+08:00 level=INFO source=server.go:1310 msg="llama runner started in 5.70 seconds"
time=2025-10-28T16:01:29.878+08:00 level=INFO source=server.go:1635 msg="llm embedding error: this model does not support embeddings"
[GIN] 2025/10/28 - 16:01:29 | 500 |    6.0015734s |             ::1 | POST     "/v1/embeddings"
time=2025-10-28T16:01:30.384+08:00 level=INFO source=server.go:1635 msg="llm embedding error: this model does not support embeddings"
[GIN] 2025/10/28 - 16:01:30 | 500 |      51.108ms |             ::1 | POST     "/v1/embeddings"
time=2025-10-28T16:01:31.315+08:00 level=INFO source=server.go:1635 msg="llm embedding error: this model does not support embeddings"
[GIN] 2025/10/28 - 16:01:31 | 500 |     48.6625ms |             ::1 | POST     "/v1/embeddings"
<!-- gh-comment-id:3455082265 --> @Imitater967 commented on GitHub (Oct 28, 2025): ``` time=2025-10-28T16:01:25.819+08:00 level=INFO source=server.go:1272 msg="waiting for llama runner to start responding" time=2025-10-28T16:01:25.819+08:00 level=INFO source=server.go:1306 msg="waiting for server to become available" status="llm server loading model" time=2025-10-28T16:01:29.857+08:00 level=INFO source=server.go:1310 msg="llama runner started in 5.70 seconds" time=2025-10-28T16:01:29.878+08:00 level=INFO source=server.go:1635 msg="llm embedding error: this model does not support embeddings" [GIN] 2025/10/28 - 16:01:29 | 500 | 6.0015734s | ::1 | POST "/v1/embeddings" time=2025-10-28T16:01:30.384+08:00 level=INFO source=server.go:1635 msg="llm embedding error: this model does not support embeddings" [GIN] 2025/10/28 - 16:01:30 | 500 | 51.108ms | ::1 | POST "/v1/embeddings" time=2025-10-28T16:01:31.315+08:00 level=INFO source=server.go:1635 msg="llm embedding error: this model does not support embeddings" [GIN] 2025/10/28 - 16:01:31 | 500 | 48.6625ms | ::1 | POST "/v1/embeddings" ```
GiteaMirror commented 2026-04-12 21:09:14 -05:00
Author
Owner
Copy Link

@Imitater967 commented on GitHub (Oct 28, 2025):

Ollama version=0.12.6 OS=Windows/10.0.26200

<!-- gh-comment-id:3455082999 --> @Imitater967 commented on GitHub (Oct 28, 2025): Ollama version=0.12.6 OS=Windows/10.0.26200
GiteaMirror commented 2026-04-12 21:09:14 -05:00
Author
Owner
Copy Link

@Durden-T commented on GitHub (Nov 7, 2025):

sames for me.
version=0.12.10
OS=windows 10

<!-- gh-comment-id:3505354415 --> @Durden-T commented on GitHub (Nov 7, 2025): sames for me. version=0.12.10 OS=windows 10
GiteaMirror commented 2026-04-12 21:09:15 -05:00
Author
Owner
Copy Link

@xzxiaoshan commented on GitHub (Nov 19, 2025):

When will this problem be solved?

<!-- gh-comment-id:3551320818 --> @xzxiaoshan commented on GitHub (Nov 19, 2025): When will this problem be solved?
GiteaMirror commented 2026-04-12 21:09:15 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Nov 19, 2025):

@xzxiaoshan

ollama run qwen3-embedding "hello there"
[0.026144993,0.012585027,-0.01587802,-0.039643414,0.004063185,-0.01620219, ... ]

Should work fine.

ollama run dengcao/Qwen3-Embedding-8B:Q4_K_M "hello there"
pulling manifest
pulling 3fcd3febec8b: 100% ▕███████████████████████████████████████████████████████████████████████████████████████████████████████████████▏ 4.7 GB
pulling 0a3d61b01340: 100% ▕███████████████████████████████████████████████████████████████████████████████████████████████████████████████▏  265 B
verifying sha256 digest
writing manifest
success
[0.026144993,0.012585027,-0.01587802,-0.039643414,0.004063185,-0.01620219,-0.021863962, ...]
<!-- gh-comment-id:3551410444 --> @pdevine commented on GitHub (Nov 19, 2025): @xzxiaoshan ``` ollama run qwen3-embedding "hello there" [0.026144993,0.012585027,-0.01587802,-0.039643414,0.004063185,-0.01620219, ... ] ``` Should work fine. ``` ollama run dengcao/Qwen3-Embedding-8B:Q4_K_M "hello there" pulling manifest pulling 3fcd3febec8b: 100% ▕███████████████████████████████████████████████████████████████████████████████████████████████████████████████▏ 4.7 GB pulling 0a3d61b01340: 100% ▕███████████████████████████████████████████████████████████████████████████████████████████████████████████████▏ 265 B verifying sha256 digest writing manifest success [0.026144993,0.012585027,-0.01587802,-0.039643414,0.004063185,-0.01620219,-0.021863962, ...] ```
GiteaMirror commented 2026-04-12 21:09:16 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Nov 19, 2025):

This is with CUDA 13 on an nvidia DGX Spark:

% OLLAMA_HOST=100.74.140.95 ./ollama run dengcao/Qwen3-Embedding-8B:Q4_K_M "hello there"
pulling manifest
pulling 3fcd3febec8b: 100% ▕███████████████████████████████████████████████████████████████████████████████████████████████████████████████▏ 4.7 GB
pulling 0a3d61b01340: 100% ▕███████████████████████████████████████████████████████████████████████████████████████████████████████████████▏  265 B
verifying sha256 digest
writing manifest
success
[0.025860516,0.012751244,-0.01591436,-0.03947152,0.004459872,-0.016639296,-0.022568734, ...]
<!-- gh-comment-id:3551424744 --> @pdevine commented on GitHub (Nov 19, 2025): This is with CUDA 13 on an nvidia DGX Spark: ``` % OLLAMA_HOST=100.74.140.95 ./ollama run dengcao/Qwen3-Embedding-8B:Q4_K_M "hello there" pulling manifest pulling 3fcd3febec8b: 100% ▕███████████████████████████████████████████████████████████████████████████████████████████████████████████████▏ 4.7 GB pulling 0a3d61b01340: 100% ▕███████████████████████████████████████████████████████████████████████████████████████████████████████████████▏ 265 B verifying sha256 digest writing manifest success [0.025860516,0.012751244,-0.01591436,-0.03947152,0.004459872,-0.016639296,-0.022568734, ...] ```
GiteaMirror commented 2026-04-12 21:09:16 -05:00
Author
Owner
Copy Link

@imbible commented on GitHub (Nov 28, 2025):

This is with CUDA 13 on an nvidia DGX Spark:

% OLLAMA_HOST=100.74.140.95 ./ollama run dengcao/Qwen3-Embedding-8B:Q4_K_M "hello there"
pulling manifest
pulling 3fcd3febec8b: 100% ▕███████████████████████████████████████████████████████████████████████████████████████████████████████████████▏ 4.7 GB
pulling 0a3d61b01340: 100% ▕███████████████████████████████████████████████████████████████████████████████████████████████████████████████▏  265 B
verifying sha256 digest
writing manifest
success
[0.025860516,0.012751244,-0.01591436,-0.03947152,0.004459872,-0.016639296,-0.022568734, ...]

It appears that you tested against chat as opposed to embedding. Did you try curl?

curl http://localhost:11434/api/embed -d '{
          "model": "hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q8_0",
          "input": "This is a test sentence to verify embedding support."
        }'

This command line spits {"error":"this model does not support embeddings"}

In addition, this contributor says the new engine does not support embedding. https://github.com/ollama/ollama/issues/10824#issuecomment-2905176025

<!-- gh-comment-id:3590627675 --> @imbible commented on GitHub (Nov 28, 2025): > This is with CUDA 13 on an nvidia DGX Spark: > > ``` > % OLLAMA_HOST=100.74.140.95 ./ollama run dengcao/Qwen3-Embedding-8B:Q4_K_M "hello there" > pulling manifest > pulling 3fcd3febec8b: 100% ▕███████████████████████████████████████████████████████████████████████████████████████████████████████████████▏ 4.7 GB > pulling 0a3d61b01340: 100% ▕███████████████████████████████████████████████████████████████████████████████████████████████████████████████▏ 265 B > verifying sha256 digest > writing manifest > success > [0.025860516,0.012751244,-0.01591436,-0.03947152,0.004459872,-0.016639296,-0.022568734, ...] > ``` It appears that you tested against chat as opposed to embedding. Did you try curl? ``` curl http://localhost:11434/api/embed -d '{ "model": "hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q8_0", "input": "This is a test sentence to verify embedding support." }' ``` This command line spits `{"error":"this model does not support embeddings"}` In addition, this contributor says the new engine does not support embedding. https://github.com/ollama/ollama/issues/10824#issuecomment-2905176025
GiteaMirror commented 2026-04-12 21:09:17 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Dec 1, 2025):

curl http://localhost:11434/api/embed -d '{
          "model": "dengcao/Qwen3-Embedding-8B:Q4_K_M",
          "input": "This is a test sentence to verify embedding support."
        }'
{"model":"dengcao/Qwen3-Embedding-8B:Q4_K_M","embeddings":[[0.017959485,-0.0114084,0.016976958,-0.01592952,0.011294606, ...

and:

% curl http://localhost:11434/api/embed -d '{
          "model": "hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q8_0",
          "input": "This is a test sentence to verify embedding support."
        }'
{"model":"hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q8_0","embeddings":[[0.017586038,-0.0084713455,0.015783766,-0.018711101,0.013251487, ...

In addition, this contributor says the new engine does not support embedding.

That's a very old comment from last May. Embbedings have been supported in the new engine for a while now.

<!-- gh-comment-id:3598011993 --> @pdevine commented on GitHub (Dec 1, 2025): ``` curl http://localhost:11434/api/embed -d '{ "model": "dengcao/Qwen3-Embedding-8B:Q4_K_M", "input": "This is a test sentence to verify embedding support." }' {"model":"dengcao/Qwen3-Embedding-8B:Q4_K_M","embeddings":[[0.017959485,-0.0114084,0.016976958,-0.01592952,0.011294606, ... ``` and: ``` % curl http://localhost:11434/api/embed -d '{ "model": "hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q8_0", "input": "This is a test sentence to verify embedding support." }' {"model":"hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q8_0","embeddings":[[0.017586038,-0.0084713455,0.015783766,-0.018711101,0.013251487, ... ``` > In addition, this contributor says the new engine does not support embedding. That's a very old comment from last May. Embbedings have been supported in the new engine for a while now.
GiteaMirror commented 2026-04-12 21:09:17 -05:00
Author
Owner
Copy Link

@imbible commented on GitHub (Dec 2, 2025):

Whatever. Still the same error on my end, and as you see in the thread, I'm not alone.

 curl http://localhost:11434/api/embed -d '{   
          "model": "hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q8_0",
          "input": "This is a test sentence to verify embedding support."
        }'
{"error":"this model does not support embeddings"}%

I'm on Ollama 0.13.0, Macbook Pro 16" M4 Max 128GB.

<!-- gh-comment-id:3599596885 --> @imbible commented on GitHub (Dec 2, 2025): Whatever. Still the same error on my end, and as you see in the thread, I'm not alone. ```sh curl http://localhost:11434/api/embed -d '{ "model": "hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q8_0", "input": "This is a test sentence to verify embedding support." }' {"error":"this model does not support embeddings"}% ``` I'm on Ollama 0.13.0, Macbook Pro 16" M4 Max 128GB.
GiteaMirror commented 2026-04-12 21:09:18 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Dec 2, 2025):

@imbible Can you share the output of ollama -v? I'm wondering if you have an older ollama server running.

<!-- gh-comment-id:3599608684 --> @pdevine commented on GitHub (Dec 2, 2025): @imbible Can you share the output of `ollama -v`? I'm wondering if you have an older ollama server running.
GiteaMirror commented 2026-04-12 21:09:18 -05:00
Author
Owner
Copy Link

@imbible commented on GitHub (Dec 2, 2025):

   ~  ollama -v  ✔  3 
ollama version is 0.13.0

<!-- gh-comment-id:3599622580 --> @imbible commented on GitHub (Dec 2, 2025):    ~  ollama -v  ✔  3  ollama version is 0.13.0
GiteaMirror commented 2026-04-12 21:09:18 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Dec 2, 2025):

Can you get the id of the model? ollama ls | grep hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q8_0
It should look like:

hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q8_0                          b818fc7f15d9    8.0 GB    26 hours ago

I'm struggling with this one as I haven't been able to duplicate it w/ either Windows or Mac.

<!-- gh-comment-id:3603768568 --> @pdevine commented on GitHub (Dec 2, 2025): Can you get the id of the model? `ollama ls | grep hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q8_0` It should look like: ``` hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q8_0 b818fc7f15d9 8.0 GB 26 hours ago ``` I'm struggling with this one as I haven't been able to duplicate it w/ either Windows or Mac.
GiteaMirror commented 2026-04-12 21:09:19 -05:00
Author
Owner
Copy Link

@imbible commented on GitHub (Dec 3, 2025):

Sure.
hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q8_0 054cb55665a3 8.0 GB 5 months ago
It appears that our hashes do not match.

<!-- gh-comment-id:3607855446 --> @imbible commented on GitHub (Dec 3, 2025): Sure. `hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q8_0 054cb55665a3 8.0 GB 5 months ago` It appears that our hashes do not match.
GiteaMirror commented 2026-04-12 21:09:19 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Dec 4, 2025):

@imbible can ollama pull hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q8_0 and see if it works? Looks like they may have updated it?

<!-- gh-comment-id:3613946070 --> @pdevine commented on GitHub (Dec 4, 2025): @imbible can `ollama pull hf.co/Qwen/Qwen3-Embedding-8B-GGUF:Q8_0` and see if it works? Looks like they may have updated it?
GiteaMirror commented 2026-04-12 21:09:20 -05:00
Author
Owner
Copy Link

@imbible commented on GitHub (Dec 5, 2025):

OK. Now it works. And the hash matches too. Interesting. The legacy version of the model definitely worked in an old version of ollama though, because I used it extensively several months ago.

<!-- gh-comment-id:3615132662 --> @imbible commented on GitHub (Dec 5, 2025): OK. Now it works. And the hash matches too. Interesting. The legacy version of the model definitely worked in an old version of ollama though, because I used it extensively several months ago.
GiteaMirror commented 2026-04-12 21:09:21 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Dec 5, 2025):

@imbible thanks for trying it out. I know we had updated the model to run on the new engine instead of the legacy engine last September. That particular model/quant is also available through ollama.com as well, although I don't know if Qwen on HF used our conversion or a different one.

I'm going to go ahead and close the issue.

<!-- gh-comment-id:3615362598 --> @pdevine commented on GitHub (Dec 5, 2025): @imbible thanks for trying it out. I know we had updated the model to run on the new engine instead of the legacy engine last September. That particular model/quant is also available through [ollama.com](https://ollama.com/library/qwen3-embedding:8b-q8_0) as well, although I don't know if Qwen on HF used our conversion or a different one. I'm going to go ahead and close the issue.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label bug needs more info
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#8461
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?