No reasoning provided by gpt-oss:20b accessed through ollama endpoint

[gianlucagilardi]

What happened?

LibreChat interface does not provide thoughts when interacting with gpt-oss:20b through ollama endpoint

Version Information

ghcr.io/danny-avila/librechat-dev latest 15d305136782 29 hours ago 1.57GB
ghcr.io/danny-avila/librechat-rag-api-dev-lite latest 33b757cfadf6 8 days ago 1.48GB
ghcr.io/danny-avila/librechat-dev e05ac102518f 9 days ago 1.17GB

Steps to Reproduce

1. Configure Remote ollama server with gpt-oss:latest (20b)
2. Configure Librechat with ollama custom endpoint

# For more information, see the Configuration Guide:
# https://www.librechat.ai/docs/configuration/librechat_yaml

# Configuration version (required)
version: 1.1.5

# Cache settings: Set to true to enable caching
#cache: true

# Definition of custom endpoints
endpoints:
  custom:
    - name: "Ollama"
      apiKey: "ollama"
      baseURL: "http://x.x.x.x:11434/v1/"
      models:
        default:
          [
            "deepseek-r1-14b:latest",
            "deepseek-r1:32b",
            "qwen2.5-14B-1M:latest",
            "qwq:latest",
            "gemma3:27b",
            "mistral-small3.1:latest",
            "qwen3:32b",
            "qwen3:30b-a3b",
            "deepseek-r1:8b",
            "qwen3:30b",
            "gpt-oss"
          ]
        fetch: false
      titleConvo: true
      titleModel: "current_model"
      summarize: false
      summaryModel: "current_model"
      forcePrompt: false
      modelDisplayLabel: "Ollama"

[...snip...]

3. In librechat select ollama/gpr-oss
4. Run any query
5. Wait for result
6. Thoughts are missing

What browsers are you seeing the problem on?

Microsoft Edge

Relevant log output

debug log:
2025-08-14T07:13:59.356Z debug: [BaseClient] Loading history:
{
  conversationId: "41703d26-d2fc-4a83-af16-3db503cddfbd",
  parentMessageId: "00000000-0000-0000-0000-000000000000",
}
2025-08-14T07:13:59.361Z debug: [BaseClient] Context Count (1/2)
{
  remainingContextTokens: 3674,
  maxContextTokens: 3686,
}
2025-08-14T07:13:59.361Z debug: [BaseClient] Context Count (2/2)
{
  remainingContextTokens: 3674,
  maxContextTokens: 3686,
}
2025-08-14T07:13:59.361Z debug: [BaseClient] tokenCountMap:
{
  88a9d622-38b2-4cb2-b03e-e69159706f82: 9,
}
2025-08-14T07:13:59.362Z debug: [BaseClient]
{
  promptTokens: 12,
  remainingContextTokens: 3674,
  payloadSize: 1,
  maxContextTokens: 3686,
}
2025-08-14T07:13:59.362Z debug: [BaseClient] tokenCountMap
{
  88a9d622-38b2-4cb2-b03e-e69159706f82: 9,
}
2025-08-14T07:13:59.362Z debug: [BaseClient] userMessage
{
  messageId: "88a9d622-38b2-4cb2-b03e-e69159706f82",
  parentMessageId: "00000000-0000-0000-0000-000000000000",
  conversationId: "41703d26-d2fc-4a83-af16-3db503cddfbd",
  sender: "User",
  text: "What is gdpr?",
  isCreatedByUser: true,
  tokenCount: 9,
}
2025-08-14T07:14:09.382Z debug: [spendTokens] conversationId: 41703d26-d2fc-4a83-af16-3db503cddfbd | Context: message | Token usage:
{
  promptTokens: 12,
  completionTokens: 261,
}
2025-08-14T07:14:09.391Z debug: [spendTokens] No transactions incurred against balance
2025-08-14T07:14:19.759Z debug: [spendTokens] conversationId: 41703d26-d2fc-4a83-af16-3db503cddfbd | Context: title | Token usage:
{
  promptTokens: undefined,
  completionTokens: undefined,
}
2025-08-14T07:14:19.759Z debug: [spendTokens] No transactions incurred against balance
2025-08-14T07:20:39.875Z debug: [BaseClient] Loading history:
{
  conversationId: "41703d26-d2fc-4a83-af16-3db503cddfbd",
  parentMessageId: "ee62bacb-1726-4dcd-ac17-8dae673c08ad",
}
2025-08-14T07:20:39.878Z debug: [BaseClient] Context Count (1/2)
{
  remainingContextTokens: 3404,
  maxContextTokens: 3686,
}
2025-08-14T07:20:39.879Z debug: [BaseClient] Context Count (2/2)
{
  remainingContextTokens: 3404,
  maxContextTokens: 3686,
}
2025-08-14T07:20:39.879Z debug: [BaseClient] tokenCountMap:
{
  88a9d622-38b2-4cb2-b03e-e69159706f82: 9,
  ee62bacb-1726-4dcd-ac17-8dae673c08ad: 261,
  b747674a-3131-433c-a009-8060c5d7195f: 9,
}
2025-08-14T07:20:39.879Z debug: [BaseClient]
{
  promptTokens: 282,
  remainingContextTokens: 3404,
  payloadSize: 3,
  maxContextTokens: 3686,
}
2025-08-14T07:20:39.880Z debug: [BaseClient] tokenCountMap
{
  88a9d622-38b2-4cb2-b03e-e69159706f82: 9,
  ee62bacb-1726-4dcd-ac17-8dae673c08ad: 261,
  b747674a-3131-433c-a009-8060c5d7195f: 9,
}
2025-08-14T07:20:39.880Z debug: [BaseClient] userMessage
{
  messageId: "b747674a-3131-433c-a009-8060c5d7195f",
  parentMessageId: "ee62bacb-1726-4dcd-ac17-8dae673c08ad",
  conversationId: "41703d26-d2fc-4a83-af16-3db503cddfbd",
  sender: "User",
  text: "what is gdpr?",
  isCreatedByUser: true,
  tokenCount: 9,
}
2025-08-14T07:20:47.172Z debug: [FinalizationRegistry] Cleaning up client for user 672351cd2aa324d483f67fd6
2025-08-14T07:20:47.173Z debug: [FinalizationRegistry] Cleaning up client for user 672351cd2aa324d483f67fd6
2025-08-14T07:21:19.753Z debug: [spendTokens] conversationId: 41703d26-d2fc-4a83-af16-3db503cddfbd | Context: message | Token usage:
{
  promptTokens: 345,
  completionTokens: 1269,
}
2025-08-14T07:21:19.774Z debug: [spendTokens] No transactions incurred against balance



No error log

Screenshots

Accessing via local cli:

Accessing via Lobehub interface:

Accessing via Librechat:

Code of Conduct

• I agree to follow this project's Code of Conduct

[danny-avila]

thanks, I see the issue here and requires some upstream changes in the agents repo and will address soon!

[reddev-aroy]

I am facing the same issue using latest lmstudio build. Seems like they changed how gpt-oss' thinking endpoint.
Below is the changelog "".

Reasoning Content from gpt-oss using the Chat Completions endpoint
This is a change in behavior compared with version 0.3.22.

message.content will no longer include reasoning content or tags.
Reasoning is now in choices.message.reasoning (non-streaming) and choices.delta.reasoning (streaming).
This matches the behavior of o3-mini.

[danny-avila]

Closed by #9085