AI Model Capability Comparison
ML engineering
Only models released within the past year are shown. Only benchmarks covering at least 3 of these models are shown. Unknown release dates are excluded; historical data is retained.
MysteryMechanismHidden mechanism inference
— means no published score for this candidate in this source, not zero.10/11 recent-year models have scores
| No. | Model | Cost / TestCost per test | Input costInput price per million tokens | Output costOutput price per million tokens | DurationReported test duration | |
|---|---|---|---|---|---|---|
| 1 |  GPT-6 AstraGPT-6 Astra | 53.15 | $1.56 | $10 | $50 | 8.15 min |
| 2 |  Claude Opus 5.5Claude Opus 5.5 | 49.55 | $4.62 | $4 | $20 | 19.63 min |
| 3 | Claude Sonnet 5.5Claude Sonnet 5.5 | 49.1 | $3.45 | $2 | $10 | 23.5 min |
| 4 | Claude Fable 5.1Claude Fable 5.1 | 47.75 | $5.63 | $10 | $50 | 17.68 min |
| 5 | GPT-6.1 SolGPT-6.1 Sol | 46.4 | $0.28 | $2 | $10 | 7.28 min |
| 6 |  Gemini 4 ArgonGemini 4 Argon | 45.49 | $4.59 | $4 | $20 | 24.92 min |
| 7 | Claude Opus 5Claude Opus 5 | 37.39 | $3.28 | $5 | $25 | 16.88 min |
| 8 | Claude Haiku 5.5Claude Haiku 5.5 | 35.13 | $0.54 | $0.1 | $0.5 | 13.42 min |
| 9 | GPT-6 SolGPT-6 Sol | 30.18 | $0.46 | $2 | $10 | 6.22 min |
| 10 | GPT-6 LunaGPT-6 Luna | 19.37 | $0.03 | $0.1 | $0.5 | 6.02 min |
PandaNpc Updated Oct 21, 2026, 06:01 UTC
Source: Epoch AI, CC BY 4.0 · Vals.ai · Artificial Analysis (artificialanalysis.ai)
Column guide
- No.
- Display order only. Sorting changes this number; it is not a published capability rank.
- Model
- The evaluated model name and its provider icon. A dash in a score column means no published score is available here, not a zero.
- Score
- Tests discovering hidden mathematical mechanisms through experiments with limited feedback. Evaluation methodology ↗
- Cost / Test
- Cost per test · USD
- Input cost
- Input price per million tokens · USD / 1M tokens
- Output cost
- Output price per million tokens · USD / 1M tokens
- Duration
- Reported test duration · s / min
- Vals.ai
- Costs and times are reported by Vals for this benchmark, not universal model prices or latency. CUA-bench estimated API cost covers evaluation usage and is a lower-bound estimate, not cost per task; time per action is an average, not a median. An asterisk marks launch-day prompt-cache behavior that increased cache-write costs. Missing values are unpublished, not zero.