← Overview
occamy-1.0
- Quantization
- NVFP4
- Reasoning
- off
- Profile
- quick
- Suite
- v2b
- Date
- 2026-09-21 – 2026-09-22
- Hardware
- NVIDIA DGX Spark, 128 GB RAM
- Inference software
- vLLM 0.26.0
- Measured
- directly at the inference server
- Overall
- 64.0
Categories
| Category | Score |
|---|
| Knowledge & reasoning | 73.1 |
| Math | 62.4 |
| Instruction following | 44.7 |
| Code | 63.0 |
| German | 76.9 |
Tasks
| Category | Task | Score | ± | n |
|---|
| Knowledge & reasoning |
mmlu_pro |
84.0 |
± 2.1 |
300 |
| Knowledge & reasoning |
gpqa_diamond_gen |
71.7 |
± 3.2 |
198 |
| Knowledge & reasoning |
musr_gen |
63.6 |
± 3.0 |
250 |
| Math |
omni_math_gen |
62.4 |
± 3.1 |
250 |
| Instruction following |
ifbench |
31.3 |
± 2.7 |
300 |
| Instruction following |
ifeval |
85.3 |
± 2.9 |
150 |
| Code |
livecodebench_gen |
63.0 |
± 3.4 |
200 |
| German |
mmlu_prox_de |
79.7 |
± 2.3 |
300 |
| German |
include_de_gen |
71.4 |
± 3.7 |
154 |
± is the standard error; the 95% interval spans roughly 1.96 standard errors around the score.
Speed
| Measure | Context (target) | Context (actual) | Median | p10 | p90 | n |
|---|
| Generation [t/s] |
– |
– |
36.7 |
36.7 |
36.7 |
5 |
| Prompt processing [t/s] |
512 |
529 |
3523.5 |
3387.4 |
3545.5 |
5 |
| Prompt processing [t/s] |
4096 |
4024 |
5539.4 |
5491.9 |
5551.9 |
5 |
| Prompt processing [t/s] |
16384 |
16001 |
5467.9 |
5449.6 |
5483.6 |
5 |
| Time to first token [ms] |
512 |
529 |
150.1 |
149.2 |
156.0 |
5 |
| Time to first token [ms] |
4096 |
4024 |
726.4 |
724.8 |
732.7 |
5 |
| Time to first token [ms] |
16384 |
16001 |
2926.3 |
2918.0 |
2936.2 |
5 |
Measured directly at the inference server.