BenchAlert
Aug 25, 2026
Sonnet 5
4.82stypical total wait
50.2 tok/s
Opus 5 completed the check 2.1× faster than GPT-5.6 Sol.
Featured
- 1Opus 55.14s46.9 tok/s+34%
- 2Grok 4.66.37s37.5 tok/s—
- 3GPT-5.6 Sol11.0s22.1 tok/s—
- –Fable 5—— tok/s—
Last 24 hours
Today’s report
Who finished fastest, what a week of each plan is worth, and the ranked board.
Quota
What 100% of a weekly limit is worth at the vendor's own API prices.
QuotaAPI $ / week
Liveupdated 6h agonext run 03:10 UTC
- Claude Max 20xB≈ $2,342
5-hour ≈ $1,013 · Fable 5 up to 50% of weekly
- ChatGPT Pro ($200, 20×)degraded≈ $2,313
- SuperGrok Heavydegraded≈ $1,627
Includes Cursor Ultra — $400/mo usage on top
- Last reading
- Aug 25
- Method
- v0.2
- Interval
- daily
Speed
Typical speed and total wait
All 10 pinned models, ranked by typical total wait over the last 24 hours. At least 12 completed checks are required.
Writing speed counts only the answer you can see. Total wait measures click-to-finish time. The table ranks the shorter total wait first.
| Rank | Model | Bar, on one shared scale | Writing speed | Total wait | vs yesterday | Answered | Coverage |
|---|---|---|---|---|---|---|---|
| 1 | Sonnet 5Anthropic · effort high | 50.22 | 4.82slikely 4.58s–5.26s | +18% | 96% | 23/24 | |
| 2 | Opus 5Anthropic · effort high | 46.89 | 5.14slikely 4.70s–5.91s | +34% | 100% | 24/24 | |
| 3 | Grok 4.5xAI · effort high | 39.43 | 6.06slikely 5.52s–6.89s | +5% | 100% | 24/24 | |
| 4 | Grok 4.6xAI · effort xhigh | 37.53 | 6.37slikely 5.99s–6.77s | +1% | 100% | 24/24 | |
| 5 | Haiku 4.5Anthropic · effort high | 38.07 | 6.42slikely 6.22s–6.91s | +26% | 92% | 22/24 | |
| 6 | GPT-5.6 TerraOpenAI · effort high | 23.24 | 10.5slikely 9.73s–11.3s | 0% | 100% | 25/24 | |
| 7 | GPT-5.6 LunaOpenAI · effort high | 22.43 | 10.9slikely 9.92s–12.1s | +8% | 100% | 24/24 | |
| 8 | GPT-5.6 SolOpenAI · effort high | 22.10 | 11.0slikely 10.00s–12.6s | -1% | 100% | 24/24 | |
| 9 | GPT-5.5OpenAI · effort high | 21.67 | 11.3slikely 10.3s–11.9s | 0% | 100% | 24/24 | |
| — | Fable 5Anthropic · effort high · collecting 10/12 · last round rate limit | — | — | — | 42% | 16/24 |
Window 24 hours · 9 of 10 models have 12 completed checks · 230 checks attempted · test kit v0.2. A dash under vs yesterday means that model did not have 12 completed checks the day before. 4 of 10 pinned models shown.
24 hours
The last 24 hours, round by round
Each line joins one model's rounds, one an hour. Every dot is a real round, so the swings stay in plain sight. Point at the plot to read any round.
Visible answer tokens per second · higher is faster
Model writing speed over time
Writing speed (tokens / second)020406080
8am2pm8pm2amnow
Aug 24Aug 25
Lab time (ET)
Line: one model's visible writing speed, check to check. Dots: the checks themselves. A single missed hour is stepped over by a faint dotted link; anything longer breaks the line. Nothing is invented to fill a gap.
Full record — every round, every number
| Round | Opus 5 | Grok 4.6 | GPT-5.6 Sol | Fable 5 |
|---|---|---|---|---|
| Aug 25 6:00am | 52.06 | 36.51 | 22.15 | — |
| Aug 25 5:00am | 49.08 | 40.88 | 24.30 | — |
| Aug 25 4:00am | 34.32 | 41.15 | 25.43 | — |
| Aug 25 3:00am | 48.09 | 36.73 | 17.67 | — |
| Aug 25 2:00am | 30.90 | 25.26 | 16.09 | — |
| Aug 25 1:00am | 37.49 | 35.70 | 24.71 | — |
| Aug 25 12:00am | 42.69 | 37.58 | 23.35 | — |
| Aug 24 11:00pm | 50.93 | 32.01 | 21.48 | — |
| Aug 24 10:00pm | 40.76 | 32.67 | 18.18 | — |
| Aug 24 9:00pm | 56.41 | 36.10 | 20.54 | — |
| Aug 24 8:00pm | 51.21 | 38.01 | 24.94 | — |
| Aug 24 7:00pm | 51.87 | 41.70 | 22.08 | — |
| Aug 24 6:00pm | 51.92 | 42.92 | 22.12 | — |
| Aug 24 5:00pm | 42.41 | 39.89 | 24.99 | — |
| Aug 24 4:00pm | 39.26 | 40.34 | 24.67 | 23.89 |
| Aug 24 3:00pm | 45.68 | 39.88 | 18.38 | 26.67 |
| Aug 24 2:00pm | 39.82 | 38.77 | 21.30 | 47.79 |
| Aug 24 1:00pm | 48.98 | 39.65 | 19.25 | 47.25 |
| Aug 24 12:00pm | 45.39 | 35.27 | 24.32 | 40.79 |
| Aug 24 11:00am | 35.71 | 21.06 | 18.43 | 43.60 |
| Aug 24 10:00am | 40.83 | 41.51 | 22.99 | 25.15 |
| Aug 24 9:00am | 51.24 | 35.32 | 23.74 | 28.07 |
| Aug 24 8:00am | 54.33 | 37.48 | 21.65 | 42.78 |
| Aug 24 7:00am | 54.44 | 31.33 | 15.70 | 42.83 |
Window 24 hours · one round every 60 minutes · 24 rounds on the clock · n = 82 completed readings drawn · test kit v0.2. 4 of 10 pinned models shown.
We also ping each app every round to check it answers at all. What that check asks lives on How we test.
How to read this
Writing speed counts only the visible answer over start-to-finish time. Everything here comes from one lab machine on test kit v0.2 — not an official vendor number.