[Bug] DeepSeek V4 Flash Responses route stalls after tool calls

[Michael-Han0608]

Thanks for the report. Include a minimal reproduction and exact environment details so we can confirm and fix the problem quickly.

Client or integration

Codex App

Area

Tools / MCP / web search

Summary

DeepSeek V4 Flash requests routed through the native Responses adapter can stall after tool calls. The upstream request returns HTTP 200 and non-zero output/reasoning tokens, but OpenCodex does not continue with the corresponding tool outputs, so no final visible answer is produced.

Plain-text turns without tools can complete normally. DeepSeek V4 Pro through the openai-chat route works normally under the same local setup.

Update: reproduced on OpenCodex 2.10.0

The same behavior remains reproducible after upgrading to OpenCodex 2.10.0:

• plain-text turns may complete normally;
• tool calls can be generated and executed;
• the upstream DeepSeek Responses request returns HTTP 200 with non-zero output/reasoning tokens;
• no corresponding function_call_output or follow-up Responses request is observed for the same turn;
• Codex App receives no final visible answer and the turn stalls.

The v2.10.0 routing change does not appear to address this continuation failure. Codex-originated requests still use the native Responses path, so the remaining defect appears to be in the Responses tool-call continuation, event handoff, or reasoning-history replay rather than initial model routing.

Reproduction

1. Use OpenCodex with the official DeepSeek API and select deepseek/deepseek-v4-flash.
2. Ensure the model is routed through the native Responses adapter.
3. Start a turn that causes the model to return one or more tool calls.
4. Observe that the tool calls may complete, but no corresponding function_call_output or follow-up request is sent for the same conversation, and no final assistant answer appears.
5. A plain-text turn without tools can complete normally. DeepSeek V4 Pro through the openai-chat route does not reproduce the problem.

The exact trigger is not yet confirmed; it may involve the native Responses event handoff, multiple tool calls, or reasoning-history replay.

Version

Originally reproduced on OpenCodex 2.9.1; reproduced again on OpenCodex 2.10.0.

Operating system

macOS

Provider and model

Official DeepSeek API / deepseek/deepseek-v4-flash

Logs or error output

Relevant local usage records:

• ocx-msbd27ow-17: HTTP 200, output tokens 568, reasoning tokens 359
• ocx-msbdeyhv-1v: HTTP 200, output tokens 401, reasoning tokens 215
• ocx-msbdfnwj-1x: HTTP 200, output tokens 113, reasoning tokens 36

The first two responses contained two completed exec_command function calls, but no corresponding function_call_output or follow-up request was observed for the same conversation.

service_tier was default; no priority, 400, or 502 error was recorded.

Screenshots and supporting files

The issue was reproduced three times after a long processing period with no visible output. Local usage records are available above.

Redacted configuration

The relevant setup uses the official DeepSeek provider, with deepseek/deepseek-v4-flash routed through the native Responses adapter. No secrets or credentials are included.

The current Responses adapter appears to sanitize reasoning.content for all Responses providers. DeepSeek's Responses API supports plain-text reasoning_text input, so this may also affect subsequent tool-call replay. This observation is not yet confirmed as the primary trigger.

Checks

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

References

• providers: route DeepSeek V4 Flash through native Responses API by default #807 — route DeepSeek V4 Flash through the native Responses API
• fix(providers): default DeepSeek V4 Flash to Responses #808 — default DeepSeek V4 Flash to Responses
• fix(responses): gate service tiers by provider capability #860 — gate service tiers by provider capability
• DeepSeek Responses API compatibility: https://api-docs.deepseek.com/guides/responses_api

[github-actions]

_{Automated translation bookkeeping — detected language: unknown.}

[github-actions]

Issue reopened

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

[Michael-Han0608]

Reproduced again on OpenCodex v2.10.0. The behavior appears unchanged from the original report:

• plain-text turns may complete normally;
• when the model emits tool calls, the calls can be generated and executed;
• the upstream DeepSeek Responses request still returns HTTP 200 with non-zero output/reasoning tokens;
• no corresponding function_call_output or follow-up Responses request is observed for the same turn;
• Codex App therefore receives no final visible answer and the turn appears to stall.

The v2.10.0 routing change does not appear to address this continuation failure. It keeps Codex-originated requests on the native Responses path, while Chat/Anthropic-originated requests use the provider-wide Chat route. The remaining defect therefore seems specific to the native Responses tool-call continuation or reasoning-history replay, rather than model selection or initial wire routing.

For reference, #808 only established deepseek-v4-flash → openai-responses as the default route and verified a comparatively simple tool loop. The failure here still appears with longer or multi-tool turns.

[lidge-jun]

Triage verdict (wt3 lane, branch codex/wt3-exec)

We split this into a local defect (fixed) and an unexplained residual (still open).

Local defect — found and fixed. On continuation requests, ocx preserved reasoning items in its continuation store (src/responses/state.ts:699, :806, :837) and through DeepSeek's stateless cleanup (src/adapters/openai-responses.ts:1003), but then sanitizeReasoningInputContent() (src/adapters/openai-responses.ts:35) blanked every non-empty reasoning item's content to [] — for EVERY Responses provider, although that rule exists only for the ChatGPT native backend. DeepSeek's Responses API accepts plaintext reasoning replay (its compatibility guide merges reasoning items into the adjacent assistant message), so every continuation reached DeepSeek with an emptied reasoning item. The fix scopes the sanitizer via a provider-level preserveResponsesReasoningContent capability (registry default on for DeepSeek; ocxr1 proxy envelopes are still stripped), with regression coverage in tests/deepseek-reasoning-replay.test.ts.

Ruled out as the cause: service_tier. DeepSeek's official Responses docs list service_tier as unsupported but state unsupported Responses parameters are silently ignored — so it cannot explain a stall. We still capability-gated it fail-closed (supportsServiceTier, PR #860's reviewed semantics) as compatibility hygiene, but that is not this bug.

Residual — not explained locally. The "no follow-up HTTP request sent at all" observation cannot be produced by any ocx code path we found: the reasoning defect only fires once a follow-up request REACHES the proxy. If no request was sent, the failure is upstream of us (client/SSE handoff) or in the serving path (external stall reports for V4-Flash are on NVIDIA NIM / vLLM, not hosted api.deepseek.com).

What would close this issue: a repro against the fixed build showing whether the follow-up request now (a) is sent and succeeds — done; (b) is sent and 400s — new evidence, send us the status + body; or (c) is still never sent — that pins the residual to the client/SSE handoff, and a capture of the client-side stream would be the next evidence we need.

Leaving open pending that confirmation; the local half ships with the wt3 branch.