Claude Opus
Claude Opus measured TTFT
Claude Opus TTFT P95 under 7s. Dated measured snapshot — not a contract SLA and not every account’s default.
- Model
- Claude Opus (catalog IDs: claude-opus-5, claude-opus-latest → Opus 5)
- Metric
- TTFT P95 — time to first byte, not first meaningful token
- Percentile
- P95
- Result
- under 7s
- Region
- Singapore
- Sample
- 10,000
- Window
- 2026-08-01 to 2026-08-07
- Network
- client network time
- Retries
- excluded
- Fallback
- excluded
Not published: input-length mix, thinking on/off, or whether the same P95 holds at peak. Model-level P50/P95 history needs a probe job writing continuously; until then this page is a method-labeled snapshot, not a live board.
Performance data comes from ModelAPI production or standardized tests. Unless an enterprise contract says otherwise, peak throughput, latency, and cache-hit rates are historical measurements or platform ceilings — they do not mean every model, region, account, or workload will sustain them. Actual performance depends on model, input length, concurrency, network location, upstream capacity, and cache policy.