"error": "Ollama is not supported. Revert to version v1.7.28 or older to use it."

[honsontran]

I'm having an issue where I have added an ollama container into the docker-compose.yml, and can confirm via curl that the web container is able to send a request to generate an output and return this back to the web container.

Also, one thing to note is that if the LLM can't be reached, the same errors requesting to downgrade will still show. Likewise, I was able to send a curl request and noticed output returning back to me from the LLM, but using the web app still resulted in the error below.

The issue I'm facing is that upon saving rules on the email assistant page, I run into this. I am currently running ollama with gpt-oss:20b:

web-1                    | [safe-action]: Calling action {
web-1                    |   "name": "saveRulesPrompt"
web-1                    | }
web-1                    | [safe-action]: Calling action {
web-1                    |   "name": "saveRulesPrompt"
web-1                    | }
web-1                    | [ai-rule]: Old prompt file {
web-1                    |   "emailAccountId": "cmel2s7110003nw0uw03wehjw",
web-1                    |   "exists": "does not exist"
web-1                    | }
web-1                    | [ai-rule]: Processing new rules prompt with AI {
web-1                    |   "emailAccountId": "cmel2s7110003nw0uw03wehjw"
web-1                    | }
web-1                    | [safe-action]: Server action error: {
web-1                    |   "metadata": {
web-1                    |     "name": "saveRulesPrompt"
web-1                    |   },
web-1                    |   "userId": "cmekza8lq0001qj0t23zve523",
web-1                    |   "userEmail": "honsontran@gmail.com",
web-1                    |   "emailAccountId": "cmel2s7110003nw0uw03wehjw",
web-1                    |   "bindArgsClientInputs": [
web-1                    |     "cmel2s7110003nw0uw03wehjw"
web-1                    |   ],
web-1                    |   "error": "Ollama is not supported. Revert to version v1.7.28 or older to use it."
web-1                    | }
web-1                    | [ai-rule]: Old prompt file {
web-1                    |   "emailAccountId": "cmel2s7110003nw0uw03wehjw",
web-1                    |   "exists": "does not exist"
web-1                    | }
web-1                    | [ai-rule]: Processing new rules prompt with AI {
web-1                    |   "emailAccountId": "cmel2s7110003nw0uw03wehjw"
web-1                    | }
web-1                    | [safe-action]: Server action error: {
web-1                    |   "metadata": {
web-1                    |     "name": "saveRulesPrompt"
web-1                    |   },
web-1                    |   "userId": "cmekza8lq0001qj0t23zve523",
web-1                    |   "userEmail": "honsontran@gmail.com",
web-1                    |   "emailAccountId": "cmel2s7110003nw0uw03wehjw",
web-1                    |   "bindArgsClientInputs": [
web-1                    |     "cmel2s7110003nw0uw03wehjw"
web-1                    |   ],
web-1                    |   "error": "Ollama is not supported. Revert to version v1.7.28 or older to use it."
web-1                    | }
web-1                    | [safe-action]: Calling action {
web-1                    |   "name": "saveRulesPrompt"
web-1                    | }
web-1                    | [safe-action]: Calling action {
web-1                    |   "name": "saveRulesPrompt"
web-1                    | }
web-1                    | [ai-rule]: Old prompt file {
web-1                    |   "emailAccountId": "cmel2s7110003nw0uw03wehjw",
web-1                    |   "exists": "does not exist"
web-1                    | }
web-1                    | [ai-rule]: Processing new rules prompt with AI {
web-1                    |   "emailAccountId": "cmel2s7110003nw0uw03wehjw"
web-1                    | }
web-1                    | [safe-action]: Server action error: {
web-1                    |   "metadata": {
web-1                    |     "name": "saveRulesPrompt"
web-1                    |   },
web-1                    |   "userId": "cmekza8lq0001qj0t23zve523",
web-1                    |   "userEmail": "honsontran@gmail.com",
web-1                    |   "emailAccountId": "cmel2s7110003nw0uw03wehjw",
web-1                    |   "bindArgsClientInputs": [
web-1                    |     "cmel2s7110003nw0uw03wehjw"
web-1                    |   ],
web-1                    |   "error": "Ollama is not supported. Revert to version v1.7.28 or older to use it."
web-1                    | }
web-1                    | [ai-rule]: Old prompt file {
web-1                    |   "emailAccountId": "cmel2s7110003nw0uw03wehjw",
web-1                    |   "exists": "does not exist"
web-1                    | }
web-1                    | [ai-rule]: Processing new rules prompt with AI {
web-1                    |   "emailAccountId": "cmel2s7110003nw0uw03wehjw"
web-1                    | }
web-1                    | [safe-action]: Server action error: {
web-1                    |   "metadata": {
web-1                    |     "name": "saveRulesPrompt"
web-1                    |   },
web-1                    |   "userId": "cmekza8lq0001qj0t23zve523",
web-1                    |   "userEmail": "honsontran@gmail.com",
web-1                    |   "emailAccountId": "cmel2s7110003nw0uw03wehjw",
web-1                    |   "bindArgsClientInputs": [
web-1                    |     "cmel2s7110003nw0uw03wehjw"
web-1                    |   ],
web-1                    |   "error": "Ollama is not supported. Revert to version v1.7.28 or older to use it."
web-1                    | }

I have also filled in these settings in my .env to access the model:

OLLAMA_BASE_URL=http://ollama:11434/api
NEXT_PUBLIC_OLLAMA_MODEL=gpt-oss:20b

ECONOMY_LLM_PROVIDER=ollama
ECONOMY_LLM_MODEL=gpt-oss:20b

# Randomly generated
INTERNAL_API_KEY=608e0fad2a5cdbfe0e437d6b8e3109ec
API_KEY_SALT=3c98e41f8c09f0d9b343770c34050784

[honsontran]

For reference, here is how I tested connectivity from the web container to the ollama container:

/app # curl http://ollama:11434/api/generate -d '{
>   "model": "gpt-oss:20b",
>   "prompt": "Hello, can you write me a haiku about Docker?"
> }'
...
...
{"model":"gpt-oss:20b","created_at":"2025-08-21T08:27:41.974899551Z","response":" safe","done":false}
{"model":"gpt-oss:20b","created_at":"2025-08-21T08:27:41.98357299Z","response":",","done":false}
{"model":"gpt-oss:20b","created_at":"2025-08-21T08:27:41.992499341Z","response":" afloat","done":false}
{"model":"gpt-oss:20b","created_at":"2025-08-21T08:27:42.0010888Z","response":".","done":false}
{"model":"gpt-oss:20b","created_at":"2025-08-21T08:27:42.00998015Z","response":"","done":true,"done_reason":"stop","context":[200006,17360,200008,3575,553,17554,162016,11,261,4410,6439,2359,22203,656,7788,17527,558,87447,100594,25,220,1323,19,12,3218,198,6576,3521,25,220,1323,20,12,3062,12,2040,279,30377,289,25,14093,279,2,13888,18403,25,8450,11,49159,11,1721,13,21030,2804,413,7360,395,1753,3176,13,200007,200006,1428,200008,13225,11,665,481,5067,668,261,2472,11169,1078,91238,30,200007,200006,173781,200005,35644,200008,976,1825,25,392,13225,11,665,481,5067,668,261,2472,11169,1078,91238,16842,3164,1682,261,2472,11169,11,261,220,20,12,22,12,20,72448,562,41339,1078,91238,13,357,1757,10635,484,13,357,2804,413,25120,1078,72448,562,43276,13,623,2472,11169,2804,413,1078,91238,13,42566,1299,1402,86277,16601,11,4207,198,18816,3024,11,22119,143924,198,54071,2461,11,7703,11,196595,364,58369,3605,72448,3232,1402,86277,350,18,8,16601,350,16,8,4207,350,16,8,871,220,20,279,18816,350,18,8,3024,350,16,936,22119,350,17,8,143924,350,16,8,871,220,18,10,16,10,17,10,16,28,22,279,54071,2461,350,17,936,7703,350,16,936,196595,350,18,8,871,220,17,10,16,10,18,28,21,30,21926,220,21,11,1309,220,20,13,17158,392,54071,2461,11,7703,11,196595,1,28991,25,38372,7,16,8,2461,7,16,936,7703,7,16,936,261,12,70969,7,17,16918,50557,196595,382,261,12,70969,25,220,17,30,2632,220,16,10,16,10,16,10,17,28,20,13,11377,196595,382,220,17,13,2632,392,54071,2461,11,7703,11,196595,1,382,220,20,13,2632,3149,2472,11169,1402,86277,16601,11,4207,198,18816,3024,11,22119,143924,198,54071,2461,11,7703,11,196595,279,13022,13,4569,5882,364,162661,11,392,133534,885,4207,11,9282,885,5604,1,5178,364,80106,316,413,5364,25,220,20,12,22,12,20,13,17291,7000,13,200007,200006,173781,200005,17196,200008,86277,16601,11,4207,4066,18816,3024,11,22119,143924,2322,4066,54071,2461,11,7703,11,196595,13],"total_duration":3416972878,"load_duration":101226992,"prompt_eval_count":79,"prompt_eval_duration":485283893,"eval_count":313,"eval_duration":2829878270}

[edgeux]

Fix: Restore Ollama Support in Inbox-Zero

Step 1: Upgrade the Provider Package

Replace the incompatible package:

cd apps/web
pnpm remove ollama-ai-provider
pnpm add ollama-ai-provider-v2

Step 2: Fix the Import

Update the import in apps/web/utils/llms/model.ts:

import { createOllama } from "ollama-ai-provider-v2";

Step 3: Restore the Ollama Case

Replace the error throw with:

case Provider.OLLAMA: {
  const modelName = aiModel || env.NEXT_PUBLIC_OLLAMA_MODEL;
  if (!modelName) throw new Error("Ollama model is not set");
  return {
    provider: Provider.OLLAMA!,
    modelName,
    model: createOllama({ baseURL: env.OLLAMA_BASE_URL })(modelName),
    backupModel: getBackupModel(aiApiKey),
  };
}

Step 4: Rebuild

rebuild blah blah

Note on Model Selection

i found Qwen3 to be the most useful