opencode-go/muse-spark-1.2-contributor 502 truncation on large tool-call tasks (mid tool call without terminal signal)

[Yuxin-Qiao]

Client or integration

Codex App

Area

Provider adapter

Summary

opencode-go/muse-spark-1.2-contributor via opencodex proxy 10100 consistently returns 502 upstream_server_error on large-context + tool-call tasks. Codex shows Reconnecting 1/5...5/5 then stream disconnected before completion: upstream stream ended mid tool call without a terminal signal — possible truncation. Small tasks succeed, large sweep always fails, deepseek-v4-flash on same task succeeds.

Expected: large tool-chain tasks should not truncate even on low, or opencodex should auto-downgrade effort / retry / fallback to a working model on 502 truncation instead of 5 ineffective retries.

Reproduction

1. opencodex 2.27.0 + codex 0.147.0, ~/.codex/config.toml model = "opencode-go/muse-spark-1.2-contributor" model_reasoning_effort = "high/medium/low" all reproduce
2. Workspace CodexBar-bug:feat, codex exec "Continue aligning the status of CodexBar-bug:feat — sweep branches, local changes and PR/CI. Just run bash tools \git status` and `git branch -vv` and summarize"`
3. localhost:10100/#logs shows 5 consecutive 502 high, ~/.opencodex/usage.jsonl ocx-mt1b32u3-d ... ocx-mt1b3tn0-l status 502 upstreamError mid tool call without terminal signal adapter openai-chat.ts:1734
4. Small tasks hi / 1+1=2 succeed with 200 on low/medium/high, switching to opencode-go/deepseek-v4-flash high on same large task succeeds 200

Version

2.27.0

Operating system

macOS 15.5 (darwin 24.x)

Provider and model

opencode-go / muse-spark-1.2-contributor

Logs or error output

requestId ocx-mt1b32u3-d ... ocx-mt1b3tn0-l
provider opencode-go model muse-spark-1.2-contributor
requestedEffort high effectiveEffort high reasoningWireValue high status 502
upstreamError: upstream stream ended mid tool call without a terminal signal — possible truncation
adapter: openai-chat src/adapters/openai-chat.ts:1734

~/.opencodex/usage.jsonl ocx-mt1aawks-4f xhigh 502 and ocx-mt1b32u3-d etc high 502
~/.opencodex/service.log no extra stack, openai-chat logs stream-truncated pendingToolCalls>0

Screenshots and supporting files

localhost:10100/#logs 2026/08/20 17:16:39-58 5x high 502 screenshots, Codex Reconnecting 5/5 screenshots available

Redacted configuration

{
  "providers": {
    "opencode-go": {
      "modelReasoningEfforts": { "muse-spark-1.2-contributor": ["low","medium","high","xhigh","max"] },
      "modelDefaultReasoningEfforts": { "muse-spark-1.2-contributor": "high" }
    }
  }
}

Checks

• I searched existing issues and documentation.
• I removed secrets, tokens, account details, request credentials, and personal data.

[coderabbitai]

⚠️ Possible Duplicate Issue(s)

• opencode-go/muse-spark-1.2: frequent 'upstream stream ended mid tool call without a terminal signal — possible truncation' in streaming tool calls (Codex App via ocx, Pi direct OK) #2156

🔗 Related PRs

#2165 - fix(openai-chat): treat a non-string repeat as padding once that field is known [merged]
#2162 - fix(anthropic): frame the opening turn so AgentRouter stops blocking non-English [merged]
#2112 - fix(catalog): allow per-provider/model opt-out of code_mode_only tool mode (#2106) [merged]
#2160 - fix(providers): identify opencode-free with the client User-Agent it claims [merged]
#2186 - [WRONG BRANCH] release: promote dev to main for 2.28.0 [closed]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!

[github-actions]

Issue reopened

The report now contains the information required by the automated check. Thanks for updating it.

[Yuxin-Qiao]

Duplicate of #2194 with comprehensive evidence, closing this one to keep single track.