Skip to content

Active Memory timeout with MiniMax-M2.7: model fails to respond within 15s timeout #66804

Description

@kumikoda110

Bug Description

Active Memory query fails with timeout when using MiniMax-M2.7 as the active model. The query (745 characters) runs for ~57 seconds then times out with no fallback, resulting in summaryChars=0.

Steps to Reproduce

  1. Configure MiniMax-M2.7 as the primary model
  2. Enable Active Memory with /verbose on or via Dreaming settings
  3. Observe that Active Memory query immediately enters timeout state

Expected Behavior

Active Memory should complete successfully, or fall back to a configured backup model if the primary model is slow.

Actual Behavior

From gateway logs:

active-memory: agent=main session=agent:main:main activeProvider=minimax activeModel=MiniMax-M2.7 start timeoutMs=15000 queryChars=745
...
embedded run failover decision: timeout
active-memory: agent=main activeModel=MiniMax-M2.7 done status=timeout elapsedMs=57071 summaryChars=0
  • timeoutMs=15000 — MiniMax times out at 15 seconds
  • elapsedMs=57071 — waited ~57 seconds total before giving up
  • summaryChars=0 — no summary returned
  • fallbackConfigured: false — no fallback model configured

Environment

  • OpenClaw: 2026.4.14 (323493f)
  • Model: minimax/MiniMax-M2.7 (MiniMax-M2.7)
  • Context window: 204800 tokens, max output 131072
  • Reasoning: enabled (reasoning=true in config)
  • Platform: macOS, Node.js 24.14.0

Configuration

{
  "models": {
    "providers": {
      "minimax": {
        "baseUrl": "https://api.minimaxi.com/anthropic",
        "api": "anthropic-messages"
      }
    }
  },
  "agents": {
    "defaults": {
      "model": {
        "primary": "minimax/MiniMax-M2.7"
      }
    }
  }
}

Analysis

From memory status --deep:

provider: "none"
searchMode: "fts-only"

Active Memory is running in fts-only mode (no vector search), which requires the model to process query results directly. The MiniMax-M2.7 model appears to timeout during this processing step.

Key observations:

  1. MiniMax-M2.7 has reasoning: true enabled — this may cause the model to enter a long reasoning loop before responding
  2. No fallback model is configured (fallbackConfigured: false)
  3. The model is called with summaryChars=0 context but still times out
  4. The 57-second total wait suggests the model eventually fails after multiple retry attempts

Workaround

None confirmed — user has no additional API keys available for fallback models.

Severity

High — Active Memory is completely unusable with MiniMax-M2.7 model.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Fields

    Priority

    None yet

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions