IncidentReliability & safetyBy DanPublished Tested Updated
DeepSeek Flash answered 200 OK and sent nothing for two hours
All reports
- ChatGPT + Codex built 33 new DMR_REMOTE toolsField testOct 23
- Gemini 4 Argon reviewed a finished episode: 14 clips, one flag, one inconsistencyField testOct 21
- Does an extra AI review step pay off? A small DeepSeek trial said noHead-to-headOct 19
- My real DeepSeek bill: 12.31 billion tokens for $122.28Cost receiptOct 17
- Gemini 4 Argon kept hanging in Gemini CLI. The cause was a 60-second timeoutIncidentOct 14
- I asked 8 AI coding agents for their biggest weaknessSelf-reportOct 14
- My agents' first game benchmark was wrong. Here's how it was caughtIncidentOct 10
- DeepSeek Flash answered 200 OK and sent nothing for two hoursIncidentSep 14
- DeepSeek V4 Flash audited 18 of my real coding sessions. Two said 'done' too earlyAuditAug 4
The report
Verdict
Upstream hang, not my setup
For about two hours the deepseek-flash endpoint accepted requests and returned zero tokens, while the provider's status page said everything was fine. Check whether the service responds outside your coding app before changing your setup.
On the tier list: A reliability record, not a ranking. It places no agent.
What it looked like
Requests to the deepseek-flash endpoint stopped producing anything. No error, no timeout from the API: requests were accepted with HTTP 200 and then no tokens arrived. From inside an agent session that looks exactly like a bug in your own setup.
How it was pinned down
Direct curl probes against the API, outside any agent harness, reproduced it. The same key answered instantly on deepseek-v4-pro, so the key, the network and the account were fine. Only deepseek-flash was hanging. When it recovered at about 21:30 UTC, responses came back in about 1.3 seconds.
The fallback was not the same model
The obvious workaround, a third-party route through OpenRouter, cost about 1.76 times as much and served third-party FP8-quantized weights, not DeepSeek's native ones. It was not quality-tested, so it is a stopgap, not a substitute.
Limits of this report
- An observational record, not a controlled benchmark.
- The outage-report count is third-party data.
- The fallback route was not quality-tested.
Receipts
DanManREAL · AI lab
Test receipt
DeepSeek V4.1 Flash
- Direct probes
- 12+ curl probes
- Outage window
- ~19:29–21:30 UTCSeptember 14, 2026, about two hours
- What the endpoint did
- HTTP 200, then zero tokens
- deepseek-v4-pro, same key, same time
- answered in under 2 s
- Response time after recovery
- ~1.3 s
- Status page during the outage
- all running smoothlyFlash shown at 99.81%
- Third-party outage reports in 24 h
- 827via StatusGator
- Fallback route cost (OpenRouter)
- ~1.76× direct$1.056 vs $0.60 per 1M output tokens, off-peak
Verdict
Upstream hang, not my setup
Tier list A reliability record, not a ranking. It places no agent.
danmanreal.com/tests/deepseek-flash-outage-2026-09-14
What was tested
On the tier list's bench
- DeepSeek V4.1 Flash · 4 receipts on file
Related
Independent testing. Sponsorships are always disclosed, and verdicts are never for sale.
Disagree? Bring your receipts