openclaw - 💡(How to fix) Fix Regression: Anthropic-compatible providers (MiniMax) truncated at ~75-130s for multi-step tool-call tasks after v2026.4.5 [1 participants]

vincentdongsheng-Dstrom · 2026-04-08T03:54:35Z

[openclaw] After upgrading from pre-v2026.4.5 was running ~v2026.3.28 to v2026.4.5 , Anthropic-compatible providers confirmed with MiniMax M2.7-highspeed via a… After upgrading from pre-v2026.4.5 (was running ~v2026.3.28) to **v2026.4.5**, Anthropic-compatible providers (confirmed with MiniMax M2.7-highspeed via `api.minimaxi.com/anthropic`) experience premature truncation for multi-step tool-call tasks at ~75-130s, regardless of `agents.defaults.timeoutSeconds` configuration. Single-turn/simple tasks work fine. ## Workaround Switched affected cron jobs from `minimax-portal/MiniMax-M2.7-highspeed` to `zai/glm-5-turbo` (OpenAI-compatible). All tasks working normally. ## Summary After upgrading from pre-v2026.4.5 (was running ~v2026.3.28) to **v2026.4.5**, Anthropic-compatible providers (confirmed with MiniMax M2.7-highspeed via `api.minimaxi.com/anthropic`) experience premature truncation for multi-step tool-call tasks at ~75-130s, regardless of `agents.defaults.timeoutSeconds` configuration. Single-turn/simple tasks work fine. ## Environment - OpenClaw: 2026.4.5 - Provider: `minimax-portal` (Anthropic-compatible endpoint) - OS: Ubuntu 24.04, systemd - `agents.defaults.timeoutSeconds: 600` (no effect on this issue) ## Observed Behavior **Before v2026.4.5:** All MiniMax cron jobs ran successfully for weeks (120-350s, 1500-6000+ output tokens per run). **After v2026.4.5:** Same jobs, same model, same prompts → truncated at ~75-130s with only ~150-900 output tokens. ### Multi-step task test (MiniMax M2.7-highspeed) | Task | Duration | Output Tokens | Result | |------|----------|---------------|--------| | Single search + summary | 18s | 362 | ✅ | | 5-step (2× search + fetch + write file + report) | 128s → timeout | 509 (truncated mid-step) | ❌ | | Cron: geopolitical news scan (multi-step) | ~75s → timeout | ~150 (truncated) | ❌ | | Cron: portfolio scan (multi-step) | ~63s → timeout | ~150 (truncated) | ❌ | ### Same tasks with OpenAI-compatible provider (GLM-5-turbo) | Task | Duration | Output Tokens | Result | |------|----------|---------------|--------| | Morning briefing (search + summarize) | 160s | 2000+ | ✅ | | Geopolitical news scan | 105s | 3000+ | ✅ | | Investment daily report | 220s | 6000+ | ✅ | Error message: `Request timed out before a response was generated. Please try again, or increase agents.defaults.timeoutSeconds in your config.` ## Attempted Fixes 1. Set `agents.defaults.timeoutSeconds: 600` → **No effect** on MiniMax/Anthropic path 2. Gateway restart after config change → **No effect** 3. Increased cron job `payload.timeoutSeconds` → **No effect** ## Hypothesis The request transport refactoring in PR #60200 ("add shared model and media request transport overrides across OpenAI-, Anthropic-, Google-, and compatible provider paths") appears to have introduced a default request-level timeout on the Anthropic-compatible transport path that: 1. Is not configurable via the public config schema (no `requestTimeoutMs` or similar field exists under `models.providers.*.request`) 2. Overrides `agents.defaults.timeoutSeconds` for Anthropic-compatible providers 3. Does not affect OpenAI-compatible providers (GLM-5-turbo works fine) ## Workaround Switched affected cron jobs from `minimax-portal/MiniMax-M2.7-highspeed` to `zai/glm-5-turbo` (OpenAI-compatible). All tasks working normally. ## Related - PR #60200 (request transport overrides) - #41475 (similar Anthropic tool payload issue, closed) - #24498 (cron scheduler timeout discrepancy)

openclaw2026-04-08 03:54:35

ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

GitHub issue URL

Helpful · Quick feedback

GitHub stats

openclaw/openclaw#62901•Fetched 2026-04-09 08:01:01

View on GitHub

Comments

Participants

Timeline

Reactions

Author

vincentdongsheng-Dstrom

Participants

vincentdongsheng-Dstrom

After upgrading from pre-v2026.4.5 (was running ~v2026.3.28) to v2026.4.5, Anthropic-compatible providers (confirmed with MiniMax M2.7-highspeed via api.minimaxi.com/anthropic) experience premature truncation for multi-step tool-call tasks at ~75-130s, regardless of agents.defaults.timeoutSeconds configuration. Single-turn/simple tasks work fine.

Error Message

Error message: Request timed out before a response was generated. Please try again, or increase agents.defaults.timeoutSeconds in your config.

Root Cause

Fix Action

Workaround

Switched affected cron jobs from minimax-portal/MiniMax-M2.7-highspeed to zai/glm-5-turbo (OpenAI-compatible). All tasks working normally.

RAW_BUFFERClick to expand / collapse

Summary

Environment

OpenClaw: 2026.4.5
Provider: minimax-portal (Anthropic-compatible endpoint)
OS: Ubuntu 24.04, systemd
agents.defaults.timeoutSeconds: 600 (no effect on this issue)

Observed Behavior

Before v2026.4.5: All MiniMax cron jobs ran successfully for weeks (120-350s, 1500-6000+ output tokens per run).

After v2026.4.5: Same jobs, same model, same prompts → truncated at ~75-130s with only ~150-900 output tokens.

Multi-step task test (MiniMax M2.7-highspeed)

Task	Duration	Output Tokens	Result
Single search + summary	18s	362	✅
5-step (2× search + fetch + write file + report)	128s → timeout	509 (truncated mid-step)	❌
Cron: geopolitical news scan (multi-step)	~75s → timeout	~150 (truncated)	❌
Cron: portfolio scan (multi-step)	~63s → timeout	~150 (truncated)	❌

Same tasks with OpenAI-compatible provider (GLM-5-turbo)

Task	Duration	Output Tokens	Result
Morning briefing (search + summarize)	160s	2000+	✅
Geopolitical news scan	105s	3000+	✅
Investment daily report	220s	6000+	✅

Error message: Request timed out before a response was generated. Please try again, or increase agents.defaults.timeoutSeconds in your config.

Attempted Fixes

Set agents.defaults.timeoutSeconds: 600 → No effect on MiniMax/Anthropic path
Gateway restart after config change → No effect
Increased cron job payload.timeoutSeconds → No effect

Hypothesis

The request transport refactoring in PR #60200 ("add shared model and media request transport overrides across OpenAI-, Anthropic-, Google-, and compatible provider paths") appears to have introduced a default request-level timeout on the Anthropic-compatible transport path that:

Is not configurable via the public config schema (no requestTimeoutMs or similar field exists under models.providers.*.request)
Overrides agents.defaults.timeoutSeconds for Anthropic-compatible providers
Does not affect OpenAI-compatible providers (GLM-5-turbo works fine)

Workaround

Switched affected cron jobs from minimax-portal/MiniMax-M2.7-highspeed to zai/glm-5-turbo (OpenAI-compatible). All tasks working normally.

PR #60200 (request transport overrides)
#41475 (similar Anthropic tool payload issue, closed)
#24498 (cron scheduler timeout discrepancy)

extent analysis

TL;DR

The most likely fix is to configure or override the default request-level timeout on the Anthropic-compatible transport path.

Guidance

Investigate the request transport refactoring in PR #60200 to understand the introduced default request-level timeout on the Anthropic-compatible transport path.
Check if there are any hidden or undocumented configuration options to override the default request-level timeout for Anthropic-compatible providers.
Consider implementing a custom transport override or patch for the Anthropic-compatible provider to increase the request-level timeout.
Verify that the agents.defaults.timeoutSeconds configuration is being applied correctly for Anthropic-compatible providers.

Example

No code snippet is provided as the issue does not contain sufficient information to create a concrete example.

Notes

The provided workaround of switching to an OpenAI-compatible provider may not be feasible for all use cases, and a more permanent solution may be required. The introduction of a default request-level timeout on the Anthropic-compatible transport path may have unintended consequences and should be thoroughly investigated.

Recommendation

Apply a workaround by configuring a custom transport override or patch for the Anthropic-compatible provider to increase the request-level timeout, as the default timeout appears to be overriding the agents.defaults.timeoutSeconds configuration. This is recommended because it allows for a more targeted solution that does not require switching to a different provider.

Vote matrix · Quick signals

Works

Did the solution work? Tap to confirm.

Easy Fix

Was it a quick fix?

Time Saver

Did it save you time?

Blocking

Was it severely blocking?

Common Issue

Are others likely hitting this too?

Flaky / Intermittent

Is it intermittent?

Verified / Reproducible

Can you reproduce it reliably?

#api #tensor shape #autograd error #model save/load #optimization

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Data

Security

Network

Code

UI/UX

Text

System

Multimedia

Protocol

API

Engineering

openclaw - 💡(How to fix) Fix Regression: Anthropic-compatible providers (MiniMax) truncated at ~75-130s for multi-step tool-call tasks after v2026.4.5 [1 participants]

Recommended Tools

GitHub issue graph ai analysis

Error Message

Root Cause

Fix Action

Workaround

Summary

Environment

Observed Behavior

Multi-step task test (MiniMax M2.7-highspeed)

Same tasks with OpenAI-compatible provider (GLM-5-turbo)

Attempted Fixes

Hypothesis

Workaround

Related

extent analysis

TL;DR

Guidance

Example

Notes

Recommendation

Still need to ship something?

TRENDING

openclaw - 💡(How to fix) Fix Regression: Anthropic-compatible providers (MiniMax) truncated at ~75-130s for multi-step tool-call tasks after v2026.4.5 [1 participants]

Recommended Tools

GitHub issue graph ai analysis

Error Message

Root Cause

Fix Action

Workaround

Summary

Environment

Observed Behavior

Multi-step task test (MiniMax M2.7-highspeed)

Same tasks with OpenAI-compatible provider (GLM-5-turbo)

Attempted Fixes

Hypothesis

Workaround

Related

extent analysis

TL;DR

Guidance

Example

Notes

Recommendation

Still need to ship something?

RELATED_DISCOVERY

TRENDING