Benchmarks · Science & research · Coding
FrontierRSI: Inference and serving
FrontierRSI's inference tasks: scheduling, routing, caching and batching for serving a frozen model with exact outputs. Scored against the best result any model reached.
9 models · 9 results · data as of 2026-11-18 · frontier-rsi.org ↗
Leaderboard
Fable 5.2 max
All results
9 rows| # | Model | Effort | Score | Harness |
|---|---|---|---|---|
| 1 | Fable 5.2 | max | 66.3% | Claude Code |
| 2 | GPT-6.1 Sol | max | 40.1% | Codex |
| 3 | Gemini 4 Carbon | high | 30.7% | Antigravity CLI |
| 4 | Opus 5.5 | max | 27.7% | Claude Code |
| 5 | Gemini 4 Argon | high | 22.9% | Antigravity CLI |
| 6 | Muse Wonder | max | 20.0% | Muse Code |
| 7 | GPT-6 Astra | max | 18.7% | Codex |
| 8 | Kimi K3.1 | max | 8.3% | Claude Code |
| 9 | Grok 4.7 | xhigh | 4.4% | Grok Build |