siyuan-note/siyuan · warning

errModelRequestTimeout

errModelRequestTimeout

Error message

model request timeout

What it means

errModelRequestTimeout is the sentinel error set when the model chat-completion request exceeds its configured requestTimeout inside createStreamWithRetry. The stream creation context is cancelled with a timer; if the provider has not returned a stream before the deadline, the agent replaces the outcome with this sentinel so classifyRetry treats it as "timeout" and applies the timeout retry/backoff policy.

Solutions

1. Increase the model request timeout in the AI provider settings so slow-to-start models get enough time.
2. Check the provider's status/latency; retry when the endpoint is responsive again (the agent retries automatically per maxRetries/classifyRetry).
3. Reduce prompt size (shorter conversation, tighter compaction) to cut time-to-first-token.
4. Verify network path (proxy/firewall) to the endpoint is not adding large latency.

Example fix

// before: requestTimeout too small for a slow reasoning model
stream, _, cancel, err := createStreamWithRetry(ctx, client, req, 2, 5*time.Second, idleTimeout, retryDelay, ch)
// after: raise the request timeout
stream, _, cancel, err := createStreamWithRetry(ctx, client, req, 2, 120*time.Second, idleTimeout, retryDelay, ch)

// before sending, sanity-check the configured timeout
if requestTimeout < 30*time.Second {
    requestTimeout = 30 * time.Second // floor for slow-to-start models
}

stream, _, cancel, err := createStreamWithRetry(ctx, client, req, maxRetries, requestTimeout, idleTimeout, retryDelay, ch)
if err != nil {
    if errors.Is(err, errModelRequestTimeout) {
        // classifyRetry returns "timeout"; rely on backoff retry or surface a timeout hint to the user
    }
}

When it happens

Trigger: In agent.go (~line 2514/2558): the OpenAI-compatible request to the model endpoint does not produce a stream within requestTimeout; requestTimedOut fires and err is set to errModelRequestTimeout, also matched later by errors.Is in classifyRetry (agent.go:2627) and in tests like TestCreateStreamRequestTimeoutAndZeroRetries.

Common situations: Slow or overloaded model provider; very long prompts (near the context limit) that take longer than the configured request timeout; network latency or VPN/proxy slowness; a requestTimeout setting that is too small for the chosen model (e.g. reasoning models with long time-to-first-token).

Understand the failure class

Background: Request timed out: what client-side request timeouts mean across libraries (Request timed out, TIMED_OUT, APITimeoutError) — this error's family across 39 libraries.

• Timeouts: ETIMEDOUT, deadlines, and hung requests — what actually expires when a request times out.

Related errors

• model request timeout
• model stream idle timeout
• agent compaction summary is empty
• agent context cannot be compacted enough
• AI editor request timeout

AI-assisted analysis of siyuan-note/siyuan@9f775e8a12 (2026-09-19). Data as JSON: /api/errors/5432ca6c028b3812. Report an issue: GitHub.

Appendix: source

Thrown at kernel/agent/agent.go:2514

			}
			e := SessionEntry{
				ID:                   id,
				Type:                 "assistant",
				Content:              m.Content,
				ReasoningCont:        m.ReasoningContent,
				ResponseOutput:       util.CloneOpenAIResponseOutput(m.ResponseOutput),
				ResponseOutputTokens: m.ResponseOutputTokens,
				RoundID:              m.RoundID,
				ToolCalls:            m.ToolCalls,
			}
			entries = append(entries, e)
		}
	}
	return entries
}

var (
	errModelRequestTimeout    = errors.New("model request timeout")
	errModelStreamIdleTimeout = errors.New("model stream idle timeout")
)

func createStreamWithRetry(ctx context.Context, client *openai.Client, req openai.ChatCompletionRequest, maxRetries int,
	requestTimeout, streamIdleTimeout time.Duration, retryDelay func(string, int) time.Duration,
	ch chan<- AgentEvent) (*util.OpenAICompletionStream, openai.ChatCompletionStreamResponse, context.CancelFunc, error) {
	return createProtocolStreamWithRetry(ctx, client, util.OpenAIProtocolChatCompletions, req, nil, maxRetries,
		requestTimeout, streamIdleTimeout, retryDelay, ch)
}

func createProtocolStreamWithRetry(ctx context.Context, client *openai.Client, protocol string,
	req openai.ChatCompletionRequest, responseInput []any, maxRetries int, requestTimeout, streamIdleTimeout time.Duration,
	retryDelay func(string, int) time.Duration,
	ch chan<- AgentEvent) (*util.OpenAICompletionStream, openai.ChatCompletionStreamResponse, context.CancelFunc, error) {
	if maxRetries < 0 {
		maxRetries = 0
	}

View on GitHub (pinned to 9f775e8a12)