~/system-one-bench / compare/fastino-glide-vs-respan-span-02
Head to head
GLiDE scores 84.1 and Span-02 79.5 on the System One Index. GLiDE is more accurate on 12 of 16 text datasets, Span-02 on 4. Same cases, same request, same scorer.
| GLiDE | Span-02 | |
|---|---|---|
| System One Index | 84.1 | 79.5 |
| English | 83.0 | 77.9 |
| Arabic | 84.8 | 82.3 |
| Vision Index | 82.5 | 86.1 |
| Calibration error | 3.6% | 10.5% |
| Median latency | – | 204 ms |
| Cost per 1,000 calls | $0.017 | $0.005 |
| GLiDE | Span-02 | |
|---|---|---|
| AG News English | 92.0 | 86.5 |
| SST-2 English | 96.0 | 95.5 |
| SST-5 English | 61.5 | 57.5 |
| DAIR Emotion English | 61.5 | 56.0 |
| Banking77 English | 91.5 | 82.0 |
| BoolQ English | 92.0 | 90.0 |
| XNLI English English | 94.7 | 83.3 |
| Typed decisions English | 75.0 | 72.1 |
| XNLI French French | 88.0 | 73.3 |
| XNLI Arabic Arabic | 83.3 | 66.0 |
| MASSIVE scenario Arabic | 73.5 | 71.0 |
| MASSIVE intent Arabic | 91.0 | 86.0 |
| OSACT4 offensive Arabic | 95.0 | 97.0 |
| Tweet sentiment Arabic | 73.5 | 76.0 |
| MIRACL rerank Arabic | 87.0 | 88.5 |
| MIRACL pairwise Arabic | 90.5 | 91.5 |
| GLiDE | Span-02 | |
|---|---|---|
| SST-2, as an image | 94.0 | 92.5 |
| Arabic sentiment, as an image | 51.5 | 70.0 |
| Oxford-IIIT Pet | 84.5 | 92.5 |
| VQAv2 yes/no | 89.0 | 87.0 |
| CountBench | 93.5 | 88.5 |
Model pages: GLiDE · Span-02. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.