RPC-Bench: leaderboard
RPC-Bench — remote-procedure-call / API protocol understanding and generation; F1-like score.
Metric: F1-like (%). Source: rpc-bench.github.io. Saturation forecast: Around January 2027. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 68.2 | #98 |
| 2 | GPT-5.2 | 65.4 | #110 |
| 3 | Claude Sonnet 4.5 | 64.64 | #142 |
| 4 | Gemini 3 Pro | 61.46 | #80 |
| 5 | Gemini 2.5 Pro | 60.79 | #152 |
| 6 | GLM-4.5 | 59.24 | #275 |
| 7 | Claude Sonnet 4 | 58.48 | #199 |
| 8 | DeepSeek V3.2 | 56.91 | #205 |
| 9 | DeepSeek V3.1 | 56.31 | #264 |
| 10 | GLM-4.7 | 53.02 | #192 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=rpc-bench · How It Works · Data refreshed daily, snapshot 2026-10-20.