AI settings cannot load model list

DEVONthink Artificial Intelligence
1

Versions:

  • DEVONthink 4.3.2 Pro Edition
  • lllama.cpp 0.42.0

I had been using this setup for a few months, but recently it has stopped working and I cannot figure out why. When I set the AI provider settings, the Model list will not load.

Screenshot 2026-09-18 at 22.10.56
Screenshot 2026-09-18 at 22.10.561772×1406 186 KB

I know the endpoint is working because I can reach it from the command line:

❯ curl -s http://localhost:9931/models | jq
{
  "data": [
    {
      "id": "ggml-org/Qwen2.5-Coder-1.5B-Instruct-Q8_0-GGUF:Q8_0",
      "aliases": [],
      "tags": [],
      "object": "model",
      "owned_by": "llamacpp",
      "created": 1789794791,
      "status": {
        "value": "unloaded",
        "args": [
          "/Users/sop/.llama-app/llama",
          "serve",
          "--host",
          "127.0.0.1",
          "--jinja",
          "--log-file",
          "/tmp/llama-server.log",
          "--port",
          "0",
          "--sleep-idle-seconds",
          "900",
          "--spec-default",
          "--alias",
          "ggml-org/Qwen2.5-Coder-1.5B-Instruct-Q8_0-GGUF:Q8_0",
          "--ctx-size",
          "4096",
          "--fit-target",
          "1024",
          "--model",
          "/Users/sop/.cache/huggingface/hub/models--ggml-org--Qwen2.5-Coder-1.5B-Instruct-Q8_0-GGUF/snapshots/9a02a57656dad824ae12a0a2d625d85e9646cc84/qwen2.5-coder-1.5b-instruct-q8_0.gguf",
          "--ubatch-size",
          "2048"
        ],
        "preset": "[ggml-org/Qwen2.5-Coder-1.5B-Instruct-Q8_0-GGUF:Q8_0]\njinja = 1\nlog-file = /tmp/llama-server.log\nsleep-idle-seconds = 900\nspec-default = 1\nctx-size = 4096\nfit-target = 1024\nmodel = /Users/sop/.cache/huggingface/hub/models--ggml-org--Qwen2.5-Coder-1.5B-Instruct-Q8_0-GGUF/snapshots/9a02a57656dad824ae12a0a2d625d85e9646cc84/qwen2.5-coder-1.5b-instruct-q8_0.gguf\nubatch-size = 2048\n\n"
      },
      "architecture": {
        "input_modalities": [
          "text"
        ],
        "output_modalities": [
          "text"
        ]
      },
      "source": "preset",
      "can_remove": false
    },
    {
      "id": "ggml-org/Qwen2.5-Coder-1.5B-Q8_0-GGUF:Q8_0",
      "aliases": [],
      "tags": [],
      "object": "model",
      "owned_by": "llamacpp",
      "created": 1789794791,
      "status": {
        "value": "unloaded",
        "args": [
          "/Users/sop/.llama-app/llama",
          "serve",
          "--host",
          "127.0.0.1",
          "--jinja",
          "--log-file",
          "/tmp/llama-server.log",
          "--port",
          "0",
          "--sleep-idle-seconds",
          "900",
          "--spec-default",
          "--alias",
          "ggml-org/Qwen2.5-Coder-1.5B-Q8_0-GGUF:Q8_0",
          "--ctx-size",
          "4096",
          "--fit-target",
          "1024",
          "--model",
          "/Users/sop/.cache/huggingface/hub/models--ggml-org--Qwen2.5-Coder-1.5B-Q8_0-GGUF/snapshots/8be1b8a895a84beea772817caaa71eba6b6e0d07/qwen2.5-coder-1.5b-q8_0.gguf",
          "--ubatch-size",
          "2048"
        ],
        "preset": "[ggml-org/Qwen2.5-Coder-1.5B-Q8_0-GGUF:Q8_0]\njinja = 1\nlog-file = /tmp/llama-server.log\nsleep-idle-seconds = 900\nspec-default = 1\nctx-size = 4096\nfit-target = 1024\nmodel = /Users/sop/.cache/huggingface/hub/models--ggml-org--Qwen2.5-Coder-1.5B-Q8_0-GGUF/snapshots/8be1b8a895a84beea772817caaa71eba6b6e0d07/qwen2.5-coder-1.5b-q8_0.gguf\nubatch-size = 2048\n\n"
      },
      "architecture": {
        "input_modalities": [
          "text"
        ],
        "output_modalities": [
          "text"
        ]
      },
      "source": "preset",
      "can_remove": false
    },
    {
      "id": "ggml-org/Qwen2.5-Coder-3B-Instruct-Q8_0-GGUF:Q8_0",
      "aliases": [],
      "tags": [],
      "object": "model",
      "owned_by": "llamacpp",
      "created": 1789794791,
      "status": {
        "value": "unloaded",
        "args": [
          "/Users/sop/.llama-app/llama",
          "serve",
          "--host",
          "127.0.0.1",
          "--jinja",
          "--log-file",
          "/tmp/llama-server.log",
          "--port",
          "0",
          "--sleep-idle-seconds",
          "900",
          "--spec-default",
          "--alias",
          "ggml-org/Qwen2.5-Coder-3B-Instruct-Q8_0-GGUF:Q8_0",
          "--ctx-size",
          "32768",
          "--fit-target",
          "1024",
          "--model",
          "/Users/sop/.cache/huggingface/hub/models--ggml-org--Qwen2.5-Coder-3B-Instruct-Q8_0-GGUF/snapshots/4944a3e9ecbaacda76873e9577d038400413772c/qwen2.5-coder-3b-instruct-q8_0.gguf",
          "--ubatch-size",
          "2048"
        ],
        "preset": "[ggml-org/Qwen2.5-Coder-3B-Instruct-Q8_0-GGUF:Q8_0]\njinja = 1\nlog-file = /tmp/llama-server.log\nsleep-idle-seconds = 900\nspec-default = 1\nctx-size = 32768\nfit-target = 1024\nmodel = /Users/sop/.cache/huggingface/hub/models--ggml-org--Qwen2.5-Coder-3B-Instruct-Q8_0-GGUF/snapshots/4944a3e9ecbaacda76873e9577d038400413772c/qwen2.5-coder-3b-instruct-q8_0.gguf\nubatch-size = 2048\n\n"
      },
      "architecture": {
        "input_modalities": [
          "text"
        ],
        "output_modalities": [
          "text"
        ]
      },
      "source": "preset",
      "can_remove": false
    },
    {
      "id": "ggml-org/Qwen2.5-Coder-3B-Q8_0-GGUF:Q8_0",
      "aliases": [],
      "tags": [],
      "object": "model",
      "owned_by": "llamacpp",
      "created": 1789794791,
      "status": {
        "value": "unloaded",
        "args": [
          "/Users/sop/.llama-app/llama",
          "serve",
          "--host",
          "127.0.0.1",
          "--jinja",
          "--log-file",
          "/tmp/llama-server.log",
          "--port",
          "0",
          "--sleep-idle-seconds",
          "900",
          "--spec-default",
          "--alias",
          "ggml-org/Qwen2.5-Coder-3B-Q8_0-GGUF:Q8_0",
          "--ctx-size",
          "4096",
          "--fit-target",
          "1024",
          "--model",
          "/Users/sop/.cache/huggingface/hub/models--ggml-org--Qwen2.5-Coder-3B-Q8_0-GGUF/snapshots/9c1de162ae417c9c3aacde97c729c4128de047d8/qwen2.5-coder-3b-q8_0.gguf",
          "--ubatch-size",
          "2048"
        ],
        "preset": "[ggml-org/Qwen2.5-Coder-3B-Q8_0-GGUF:Q8_0]\njinja = 1\nlog-file = /tmp/llama-server.log\nsleep-idle-seconds = 900\nspec-default = 1\nctx-size = 4096\nfit-target = 1024\nmodel = /Users/sop/.cache/huggingface/hub/models--ggml-org--Qwen2.5-Coder-3B-Q8_0-GGUF/snapshots/9c1de162ae417c9c3aacde97c729c4128de047d8/qwen2.5-coder-3b-q8_0.gguf\nubatch-size = 2048\n\n"
      },
      "architecture": {
        "input_modalities": [
          "text"
        ],
        "output_modalities": [
          "text"
        ]
      },
      "source": "preset",
      "can_remove": false
    },
    {
      "id": "ggml-org/Qwen3.8-27B-GGUF:Q8_0",
      "aliases": [],
      "tags": [],
      "object": "model",
      "owned_by": "llamacpp",
      "created": 1789794791,
      "status": {
        "value": "unloaded",
        "args": [
          "/Users/sop/.llama-app/llama",
          "serve",
          "--host",
          "127.0.0.1",
          "--jinja",
          "--log-file",
          "/tmp/llama-server.log",
          "--port",
          "0",
          "--sleep-idle-seconds",
          "900",
          "--spec-default",
          "--model-draft",
          "/Users/sop/.cache/huggingface/hub/models--ggml-org--Qwen3.8-27B-GGUF/snapshots/0669b98607d47046c7c2b3f801011d54a08cfccf/mtp-Qwen3.8-27B-Q8_0.gguf",
          "--spec-draft-n-max",
          "3",
          "--spec-type",
          "draft-mtp",
          "--alias",
          "ggml-org/Qwen3.8-27B-GGUF:Q8_0",
          "--ctx-size",
          "32768",
          "--fit-target",
          "1024",
          "--model",
          "/Users/sop/.cache/huggingface/hub/models--ggml-org--Qwen3.8-27B-GGUF/snapshots/0669b98607d47046c7c2b3f801011d54a08cfccf/Qwen3.8-27B-Q8_0.gguf",
          "--mmproj",
          "/Users/sop/.cache/huggingface/hub/models--ggml-org--Qwen3.8-27B-GGUF/snapshots/0669b98607d47046c7c2b3f801011d54a08cfccf/mmproj-Qwen3.8-27B-Q8_0.gguf",
          "--ubatch-size",
          "2048"
        ],
        "preset": "[ggml-org/Qwen3.8-27B-GGUF:Q8_0]\njinja = 1\nlog-file = /tmp/llama-server.log\nsleep-idle-seconds = 900\nspec-default = 1\nmodel-draft = /Users/sop/.cache/huggingface/hub/models--ggml-org--Qwen3.8-27B-GGUF/snapshots/0669b98607d47046c7c2b3f801011d54a08cfccf/mtp-Qwen3.8-27B-Q8_0.gguf\nspec-draft-n-max = 3\nspec-type = draft-mtp\nctx-size = 32768\nfit-target = 1024\nmodel = /Users/sop/.cache/huggingface/hub/models--ggml-org--Qwen3.8-27B-GGUF/snapshots/0669b98607d47046c7c2b3f801011d54a08cfccf/Qwen3.8-27B-Q8_0.gguf\nmmproj = /Users/sop/.cache/huggingface/hub/models--ggml-org--Qwen3.8-27B-GGUF/snapshots/0669b98607d47046c7c2b3f801011d54a08cfccf/mmproj-Qwen3.8-27B-Q8_0.gguf\nubatch-size = 2048\n\n"
      },
      "architecture": {
        "input_modalities": [
          "text",
          "image"
        ],
        "output_modalities": [
          "text"
        ]
      },
      "source": "preset",
      "can_remove": false
    },
    {
      "id": "ggml-org/gemma-4-12B-it-GGUF:Q8_0",
      "aliases": [],
      "tags": [],
      "object": "model",
      "owned_by": "llamacpp",
      "created": 1789794791,
      "status": {
        "value": "loaded",
        "args": [
          "/Users/sop/.llama-app/llama",
          "serve",
          "--host",
          "127.0.0.1",
          "--jinja",
          "--log-file",
          "/tmp/llama-server.log",
          "--port",
          "64962",
          "--sleep-idle-seconds",
          "900",
          "--spec-default",
          "--model-draft",
          "/Users/sop/.cache/huggingface/hub/models--ggml-org--gemma-4-12B-it-GGUF/snapshots/7e0fbb8205d1f4857f4606a38a65023aaeb5f544/mtp-gemma-4-12B-it-Q8_0.gguf",
          "--spec-draft-n-max",
          "3",
          "--spec-type",
          "draft-mtp",
          "--alias",
          "ggml-org/gemma-4-12B-it-GGUF:Q8_0",
          "--ctx-size",
          "131072",
          "--fit-target",
          "1024",
          "--model",
          "/Users/sop/.cache/huggingface/hub/models--ggml-org--gemma-4-12B-it-GGUF/snapshots/7e0fbb8205d1f4857f4606a38a65023aaeb5f544/gemma-4-12B-it-Q8_0.gguf",
          "--mmproj",
          "/Users/sop/.cache/huggingface/hub/models--ggml-org--gemma-4-12B-it-GGUF/snapshots/7e0fbb8205d1f4857f4606a38a65023aaeb5f544/mmproj-gemma-4-12B-it-Q8_0.gguf",
          "--ubatch-size",
          "2048"
        ],
        "preset": "[ggml-org/gemma-4-12B-it-GGUF:Q8_0]\njinja = 1\nlog-file = /tmp/llama-server.log\nsleep-idle-seconds = 900\nspec-default = 1\nmodel-draft = /Users/sop/.cache/huggingface/hub/models--ggml-org--gemma-4-12B-it-GGUF/snapshots/7e0fbb8205d1f4857f4606a38a65023aaeb5f544/mtp-gemma-4-12B-it-Q8_0.gguf\nspec-draft-n-max = 3\nspec-type = draft-mtp\nctx-size = 131072\nfit-target = 1024\nmodel = /Users/sop/.cache/huggingface/hub/models--ggml-org--gemma-4-12B-it-GGUF/snapshots/7e0fbb8205d1f4857f4606a38a65023aaeb5f544/gemma-4-12B-it-Q8_0.gguf\nmmproj = /Users/sop/.cache/huggingface/hub/models--ggml-org--gemma-4-12B-it-GGUF/snapshots/7e0fbb8205d1f4857f4606a38a65023aaeb5f544/mmproj-gemma-4-12B-it-Q8_0.gguf\nubatch-size = 2048\n\n"
      },
      "architecture": {
        "input_modalities": [
          "text",
          "image",
          "audio"
        ],
        "output_modalities": [
          "text"
        ]
      },
      "source": "preset",
      "can_remove": false,
      "meta": {
        "vocab_type": true,
        "n_vocab": 262144,
        "n_ctx": 131072,
        "n_ctx_train": 262144,
        "n_embd": 3840,
        "n_params": 11907350576,
        "size": 12653822144,
        "ftype": "Q8_0"
      }
    },
    {
      "id": "ggml-org/gemma-4-26B-A4B-it-GGUF:Q8_0",
      "aliases": [],
      "tags": [],
      "object": "model",
      "owned_by": "llamacpp",
      "created": 1789794791,
      "status": {
        "value": "unloaded",
        "args": [
          "/Users/sop/.llama-app/llama",
          "serve",
          "--host",
          "127.0.0.1",
          "--jinja",
          "--log-file",
          "/tmp/llama-server.log",
          "--port",
          "0",
          "--sleep-idle-seconds",
          "900",
          "--spec-default",
          "--model-draft",
          "/Users/sop/.cache/huggingface/hub/models--ggml-org--gemma-4-26B-A4B-it-GGUF/snapshots/bb4531cda34d1ea09d9814959ed4d5833cf2a4c8/mtp-gemma-4-26B-A4B-it-Q8_0.gguf",
          "--spec-draft-n-max",
          "3",
          "--spec-type",
          "draft-mtp",
          "--alias",
          "ggml-org/gemma-4-26B-A4B-it-GGUF:Q8_0",
          "--ctx-size",
          "65536",
          "--fit-target",
          "1024",
          "--model",
          "/Users/sop/.cache/huggingface/hub/models--ggml-org--gemma-4-26B-A4B-it-GGUF/snapshots/bb4531cda34d1ea09d9814959ed4d5833cf2a4c8/gemma-4-26B-A4B-it-Q8_0.gguf",
          "--mmproj",
          "/Users/sop/.cache/huggingface/hub/models--ggml-org--gemma-4-26B-A4B-it-GGUF/snapshots/bb4531cda34d1ea09d9814959ed4d5833cf2a4c8/mmproj-gemma-4-26B-A4B-it-Q8_0.gguf",
          "--ubatch-size",
          "2048"
        ],
        "preset": "[ggml-org/gemma-4-26B-A4B-it-GGUF:Q8_0]\njinja = 1\nlog-file = /tmp/llama-server.log\nsleep-idle-seconds = 900\nspec-default = 1\nmodel-draft = /Users/sop/.cache/huggingface/hub/models--ggml-org--gemma-4-26B-A4B-it-GGUF/snapshots/bb4531cda34d1ea09d9814959ed4d5833cf2a4c8/mtp-gemma-4-26B-A4B-it-Q8_0.gguf\nspec-draft-n-max = 3\nspec-type = draft-mtp\nctx-size = 65536\nfit-target = 1024\nmodel = /Users/sop/.cache/huggingface/hub/models--ggml-org--gemma-4-26B-A4B-it-GGUF/snapshots/bb4531cda34d1ea09d9814959ed4d5833cf2a4c8/gemma-4-26B-A4B-it-Q8_0.gguf\nmmproj = /Users/sop/.cache/huggingface/hub/models--ggml-org--gemma-4-26B-A4B-it-GGUF/snapshots/bb4531cda34d1ea09d9814959ed4d5833cf2a4c8/mmproj-gemma-4-26B-A4B-it-Q8_0.gguf\nubatch-size = 2048\n\n"
      },
      "architecture": {
        "input_modalities": [
          "text",
          "image"
        ],
        "output_modalities": [
          "text"
        ]
      },
      "source": "preset",
      "can_remove": false
    },
    {
      "id": "ggml-org/gpt-oss-20b-GGUF:MXFP4",
      "aliases": [],
      "tags": [],
      "object": "model",
      "owned_by": "llamacpp",
      "created": 1789794791,
      "status": {
        "value": "unloaded",
        "args": [
          "/Users/sop/.llama-app/llama",
          "serve",
          "--host",
          "127.0.0.1",
          "--jinja",
          "--log-file",
          "/tmp/llama-server.log",
          "--port",
          "0",
          "--sleep-idle-seconds",
          "900",
          "--spec-default",
          "--alias",
          "ggml-org/gpt-oss-20b-GGUF:MXFP4",
          "--ctx-size",
          "131072",
          "--fit-target",
          "1024",
          "--model",
          "/Users/sop/.cache/huggingface/hub/models--ggml-org--gpt-oss-20b-GGUF/snapshots/ef9b12f2ff56c69cf32153a02784e7a3c88bf524/gpt-oss-20b-MXFP4.gguf",
          "--ubatch-size",
          "2048"
        ],
        "preset": "[ggml-org/gpt-oss-20b-GGUF:MXFP4]\njinja = 1\nlog-file = /tmp/llama-server.log\nsleep-idle-seconds = 900\nspec-default = 1\nctx-size = 131072\nfit-target = 1024\nmodel = /Users/sop/.cache/huggingface/hub/models--ggml-org--gpt-oss-20b-GGUF/snapshots/ef9b12f2ff56c69cf32153a02784e7a3c88bf524/gpt-oss-20b-MXFP4.gguf\nubatch-size = 2048\n\n"
      },
      "architecture": {
        "input_modalities": [
          "text"
        ],
        "output_modalities": [
          "text"
        ]
      },
      "source": "preset",
      "can_remove": false
    }
  ],
  "object": "list"
}

In ~/Library/Application Support/DEVONthink/Chat.log I see these messages:

Loading of models failed: Invalid HTTP status 404

(it seems like 1 line like that for each model).

How can I troubleshoot this? I can’t see what request DEVONthink is trying to make when getting this HTTP 404 status response.

2

The URL is incomplete, the suffix /chat/completions is required. In this case http://127.0.0.1:9931/v1/chat/completions.

The next maintenance release will show an error message in case of incomplete/invalid URLs.

1 Like
3

Thank you @cgrunenberg!

However, that only partially works.

I only see 2 out of my 8 models in the list:

Screenshot 2026-09-19 at 10.27.09
Screenshot 2026-09-19 at 10.27.09950×176 42.3 KB

The Chat.log shows this when I hit the reload models button:

2026-09-19 10:28:14,477 ERROR: Loading of model info failed: Invalid HTTP status 404
2026-09-19 10:28:14,478 ERROR: Loading of model info failed: Invalid HTTP status 404
2026-09-19 10:28:14,479 ERROR: Loading of model info failed: Invalid HTTP status 404
2026-09-19 10:28:14,479 ERROR: Loading of model info failed: Invalid HTTP status 404
2026-09-19 10:28:14,479 ERROR: Loading of model info failed: Invalid HTTP status 404
2026-09-19 10:28:14,479 ERROR: Loading of model info failed: Invalid HTTP status 404
2026-09-19 10:28:14,480 ERROR: Loading of model info failed: Invalid HTTP status 404
2026-09-19 10:28:14,480 ERROR: Loading of model info failed: Invalid HTTP status 404
2026-09-19 10:28:14,491 ERROR: Loading of models failed: Invalid HTTP status 404
4

Does reloading the model list and/or relaunching the app make a difference? And just curious but what’s the use case for these tiny and old Qwen 2.5 Coder models?

5

Unfortunately neither reloading the model list or relaunching the app makes a difference.

These Qwen 2.5 Coder models got installed as defaults with the Llama.cpp VSCode plugin, but I don’t particularly need them. Actually I just deleted them to see if that would help and it did not.

The model that was working before was ggml-org/gemma-4-12B-it-GGUF:Q8_0, this one works fairly well for file renaming which is mainly what I’m trying to do.

6

The next release will include some improvements regarding model probing. However, for local processing Ollama or LM Studio are highly recommended as they do not need this.

1 Like
7

Thanks, I’ve tried both in the past but llama.cpp seems much faster than either one of those. I found a workaround though, I can start up an individual llama.cpp server for a single model and it seems to work

> llama-server -m ${GGUF_FILE}

...
llama_server: listening on http://127.0.0.1:8080

So I can set the endpoint to that now and it works.