~/system-one-bench / compare/baseten-elephant-1-vs-openai-gpt-6-luna-decisions
Head to head
Elephant 1 scores 82.1 and GPT-6 Luna Decisions 76.0 on the System One Index. Elephant 1 is more accurate on 12 of 16 text datasets, GPT-6 Luna Decisions on 4. Same cases, same request, same scorer.
| Elephant 1 | GPT-6 Luna Decisions | |
|---|---|---|
| System One Index | 82.1 | 76.0 |
| English | 81.3 | 75.2 |
| Arabic | 82.0 | 78.8 |
| Vision Index | – | 81.6 |
| Calibration error | 3.7% | 9.8% |
| Median latency | – | 215 ms |
| Cost per 1,000 calls | $0.013 | $0.040 |
| Elephant 1 | GPT-6 Luna Decisions | |
|---|---|---|
| AG News English | 89.0 | 87.5 |
| SST-2 English | 94.5 | 93.5 |
| SST-5 English | 58.5 | 61.5 |
| DAIR Emotion English | 54.5 | 58.5 |
| Banking77 English | 94.5 | 77.5 |
| BoolQ English | 91.0 | 85.5 |
| XNLI English English | 94.0 | 66.7 |
| Typed decisions English | 74.2 | 71.0 |
| XNLI French French | 90.0 | 62.0 |
| XNLI Arabic Arabic | 79.3 | 54.0 |
| MASSIVE scenario Arabic | 69.0 | 66.0 |
| MASSIVE intent Arabic | 89.5 | 84.5 |
| OSACT4 offensive Arabic | 91.0 | 93.5 |
| Tweet sentiment Arabic | 66.0 | 80.0 |
| MIRACL rerank Arabic | 88.0 | 86.0 |
| MIRACL pairwise Arabic | 91.0 | 87.5 |
| Elephant 1 | GPT-6 Luna Decisions | |
|---|---|---|
| SST-2, as an image | – | 88.5 |
| Arabic sentiment, as an image | – | 67.0 |
| Oxford-IIIT Pet | – | 87.0 |
| VQAv2 yes/no | – | 80.5 |
| CountBench | – | 85.0 |
Model pages: Elephant 1 · GPT-6 Luna Decisions. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.