Model rankings
Built from 15 eligible real-world runs. Under-review runs are excluded. Data as of Nov 4, 2026, 20:00 UTC. How we score
Filters & weights1 active
-AClaude Sonnet 5.5default · thinking off(fewer than 5 runs)
78 (too few runs for a range)Pass rate: 100%(1/1)
- Cost / success
- $0.06
- Avg time
- 44s
- Step-ins
- 2.0
1 run · first-try 100%Compare runs →
-KKimi K3.1 Pro (Venice)thinking on(fewer than 5 runs)
70 (too few runs for a range)Pass rate: 100%(1/1)
- Cost / success
- $0.15
- Avg time
- 2m
- Step-ins
- 2.0
1 run · first-try 100%Compare runs →
-AClaude Opus 5.5default · thinking on(fewer than 5 runs)
66.5 (44.4–88.7)Pass rate: 67%(2/3, 1 partial)
- Cost / success
- $0.30
- Avg time
- 35s
- Step-ins
- 1.7
3 runs · first-try 67%Compare runs →
-AClaude Haiku 5.5default · thinking on(fewer than 5 runs)
51.6 (24–79.1)Pass rate: 33%(1/3, 1 partial, 1 fail)
- Cost / success
- $0.02
- Avg time
- 37s
- Step-ins
- 1.0
3 runs · first-try 33%Compare runs →
-AClaude Sonnet 5.5default · thinking on(fewer than 5 runs)
49.6 (8.4–90.8)Pass rate: 50%(1/2, 1 fail)
- Cost / success
- $0.08
- Avg time
- 31s
- Step-ins
- 1.0
2 runs · first-try 50%Compare runs →
-MLlama 3.2 3B (local)thinking off(fewer than 5 runs)
25 (too few runs for a range)Pass rate: 0%(0/1, 1 fail)
- Cost / success
- -
- Avg time
- 10s
- Step-ins
- 0.0
1 run · first-try 0%Compare runs →
-QQwen3 Coder 480B Turbo (Venice)thinking on(fewer than 5 runs)
25 (too few runs for a range)Pass rate: 0%(0/1, 1 fail)
- Cost / success
- -
- Avg time
- 38s
- Step-ins
- 0.0
1 run · first-try 0%Compare runs →
-MMistral Small 3.2 24B Instruct (Venice)thinking on(fewer than 5 runs)
17.1 (7.8–26.5)Pass rate: 0%(0/2, 2 fail)
- Cost / success
- -
- Avg time
- 23s
- Step-ins
- 1.0
2 runs · first-try 0%Compare runs →
-QQwen2.5 7B (local)thinking off(fewer than 5 runs)
15 (too few runs for a range)Pass rate: 0%(0/1, 1 fail)
- Cost / success
- -
- Avg time
- 4m
- Step-ins
- 0.0
1 run · first-try 0%Compare runs →
Not enough data to rank yet: a model needs at least 5 eligible runs here before it gets a rank number. The models below are listed with their real results so far. Add a run to help fill this in.
Want a side-by-side? Click a model, tick Compare on 2–4 runs, then open the tray at the bottom. How compare works
What the columns mean
- Pass rate: share of runs that met 100% of the task's success criteria. Partial and failed runs are counted and shown, not hidden.
- Step-ins: how many times a person (or a script) had to jump in after the first prompt to correct, clarify, approve or nudge the AI. 0 means it finished on its own. Most runs here come from automated harnesses that send one prompt and never intervene, so 0 step-ins is a measured value, not a placeholder.
- Trust: how much we believe a run's numbers, from automatic checks (evidence log, recomputed cost, consistency). Automated runs with a full log and no warnings typically land at 80%.
- Trend arrows (rank change vs 7 days ago) appear once there is enough history to compare; none are shown yet.
Head-to-head column hidden: it needs 20+ rerun pairs, and there aren't enough yet.
How rankings work
- Each run is scored 0–100 from quality, success (incl. first-try), cost-per-success and speed (percentile vs. other runs of the same use case), and autonomy, using the weights above.
- Runs are weighted by trust score and recency (half-life 90 days). The range shown is a 90% confidence interval; the whisker on each bar spans it. Models with too few runs for a real interval show no range.
- Models with fewer than 5 runs are listed but not numbered. Head-to-head reruns feed a Bradley-Terry rating, blended at 30% once there are 20+ pairs.
- Under-review runs never count. Sponsorship never affects rankings. Full details: Methodology.