~/system-one-bench / compare/fastino-glide-vs-nace-ai-drex-v1-5
Head to head
GLiDE scores 84.1 and Drex v1.5 77.6 on the System One Index. GLiDE is more accurate on 15 of 16 text datasets, Drex v1.5 on 1. Same cases, same request, same scorer.
| GLiDE | Drex v1.5 | |
|---|---|---|
| System One Index | 84.1 | 77.6 |
| English | 83.0 | 77.6 |
| Arabic | 84.8 | 77.0 |
| Vision Index | 82.5 | – |
| Calibration error | 3.6% | 10.8% |
| Median latency | – | 198 ms |
| Cost per 1,000 calls | $0.017 | $0.010 |
| GLiDE | Drex v1.5 | |
|---|---|---|
| AG News English | 92.0 | 90.0 |
| SST-2 English | 96.0 | 94.5 |
| SST-5 English | 61.5 | 53.5 |
| DAIR Emotion English | 61.5 | 53.5 |
| Banking77 English | 91.5 | 87.0 |
| BoolQ English | 92.0 | 89.0 |
| XNLI English English | 94.7 | 88.7 |
| Typed decisions English | 75.0 | 64.8 |
| XNLI French French | 88.0 | 82.0 |
| XNLI Arabic Arabic | 83.3 | 74.0 |
| MASSIVE scenario Arabic | 73.5 | 58.5 |
| MASSIVE intent Arabic | 91.0 | 77.5 |
| OSACT4 offensive Arabic | 95.0 | 90.5 |
| Tweet sentiment Arabic | 73.5 | 63.5 |
| MIRACL rerank Arabic | 87.0 | 89.0 |
| MIRACL pairwise Arabic | 90.5 | 86.0 |
| GLiDE | Drex v1.5 | |
|---|---|---|
| SST-2, as an image | 94.0 | – |
| Arabic sentiment, as an image | 51.5 | – |
| Oxford-IIIT Pet | 84.5 | – |
| VQAv2 yes/no | 89.0 | – |
| CountBench | 93.5 | – |
Model pages: GLiDE · Drex v1.5. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.