~/system-one-bench / compare/baseten-elephant-1-vs-cloudflare-clef-omni
Head to head
Elephant 1 scores 82.1 and Clef Omni 79.7 on the System One Index. Elephant 1 is more accurate on 10 of 16 text datasets, Clef Omni on 5, and 1 are tied. Same cases, same request, same scorer.
| Elephant 1 | Clef Omni | |
|---|---|---|
| System One Index | 82.1 | 79.7 |
| English | 81.3 | 78.6 |
| Arabic | 82.0 | 80.5 |
| Vision Index | – | 72.2 |
| Calibration error | 3.7% | 3.4% |
| Median latency | – | 208 ms |
| Cost per 1,000 calls | $0.013 | $0.074 |
| Elephant 1 | Clef Omni | |
|---|---|---|
| AG News English | 89.0 | 89.0 |
| SST-2 English | 94.5 | 97.0 |
| SST-5 English | 58.5 | 51.0 |
| DAIR Emotion English | 54.5 | 56.0 |
| Banking77 English | 94.5 | 96.0 |
| BoolQ English | 91.0 | 90.0 |
| XNLI English English | 94.0 | 82.7 |
| Typed decisions English | 74.2 | 67.0 |
| XNLI French French | 90.0 | 82.7 |
| XNLI Arabic Arabic | 79.3 | 73.3 |
| MASSIVE scenario Arabic | 69.0 | 89.0 |
| MASSIVE intent Arabic | 89.5 | 91.5 |
| OSACT4 offensive Arabic | 91.0 | 90.5 |
| Tweet sentiment Arabic | 66.0 | 50.0 |
| MIRACL rerank Arabic | 88.0 | 83.0 |
| MIRACL pairwise Arabic | 91.0 | 86.5 |
| Elephant 1 | Clef Omni | |
|---|---|---|
| SST-2, as an image | – | 89.5 |
| Arabic sentiment, as an image | – | 28.5 |
| Oxford-IIIT Pet | – | 88.0 |
| VQAv2 yes/no | – | 87.0 |
| CountBench | – | 68.0 |
Model pages: Elephant 1 · Clef Omni. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.