Benchmarks · agentic
Time Horizon Index: KSP
Progress through 30 Kerbal Space Program missions during a five-day autonomous run, with partial credit within a mission.
13 models · 13 results · updated 2026-10-07 · vals.ai ↗
Leaderboard
Opus 5.5 max
All results
13 rows| # | Model | Effort | Time Horizon Index: KSP | Subscription $ / task | API $ / task |
|---|---|---|---|---|---|
| 1 | Opus 5.5 | max | 91.3% | $20 | $1303 |
| 2 | GPT-6 Astra | max | 90.5% | $360 | $4207 |
| 3 | Fable 5.1 | max | 63.3% | $726 | $4655 |
| 4 | Gemini 4 Argon | high | 52.5% | $253 | $709 |
| 5 | GPT-5.6 Sol | max | 23.8% | $112 | $2027 |
| 6 | Opus 5 | max | 18.8% | $23 | $1349 |
| 7 | Gemini 3.8 Flash | high | 18.8% | $22 | $543 |
| 8 | Opus 4.8 | max | 11.8% | $26 | $1499 |
| 9 | Kimi K3 | max | 10.5% | $17 | $216 |
| 10 | GPT-5.5 | xhigh | 8.3% | $29 | $516 |
| 11 | Grok 4.5 | high | 6.3% | $54 | $344 |
| 12 | Grok 4.6 | high | 5.8% | $23 | $225 |
| 13 | Muse Spark 1.1 | xhigh | 1.7% | $145 | $145 |