Compare / Step3-VL-10B vs Apertus 1.5 8B
Step3-VL-10B vs Apertus 1.5 8B
Updated November 2026
Verdict
Step3-VL-10B scores higher on the Small LM Index: 53.2 vs 23.5. On the 1 benchmark both labs report, Step3-VL-10B wins 1 and Apertus 1.5 8B wins 0. Both need about 8 GB of memory at 4-bit quantization. Neither model has a hosted API on OpenRouter, so you would run both yourself.
Pick Step3-VL-10B if…
- you want the strongest model overall
Pick Apertus 1.5 8B if…
- you work with long documents
- you want the more widely used model (more quantized builds and community support)
Specifications
| Step3-VL-10B | Apertus 1.5 8B | |
|---|---|---|
| Small LM Index | 53.2 (likely 40–64) | 23.5 (likely 16–34) |
| Lab | StepFun · China | Swiss AI Initiative (Apertus) · Europe |
| Released | 2026-01 | 2026-07-24 |
| Parameters | 10B (dense) | 8.9B (dense) |
| Memory at 4-bit | ~8 GB | ~7.2 GB |
| Context window | 64K tokens | 262K tokens |
| Input | text, image | text, image, audio |
| Reasoning | Always thinks | Hybrid (thinking on/off) |
| License | Apache-2.0 | Apache-2.0 |
| API price / 1M tokens | Self-host only | Self-host only |
| Hugging Face downloads (30 days) | — | 22.8K |
Benchmarks
Higher is better. Bold marks the winner where both labs report a score. ≈ marks the index model's estimate for a score that wasn't reported, ✓ an independent measurement and * a score measured by another lab.
| Benchmark | Step3-VL-10B | Apertus 1.5 8B |
|---|---|---|
| MMLU-Pro Knowledge | ≈ 73 | 48.8 |
| GPQA Diamond Knowledge | 69.0✓ | 27.2 |
| Humanity's Last Exam Reasoning | 10.8✓ | ≈ 4 |
| τ²-Bench Agentic | 16.1✓ | — |
| IFBench Instruction following | 50.2✓ | — |
| AA-LCR Long context | 0.0✓ | — |
Scores are mostly reported by each lab with its own settings, so small gaps are noise. See the methodology. Open in the interactive comparison to add more models.
Questions
Which is better, Step3-VL-10B or Apertus 1.5 8B?
Step3-VL-10B scores higher on the Small LM Index: 53.2 vs 23.5. On the 1 benchmark both labs report, Step3-VL-10B wins 1 and Apertus 1.5 8B wins 0.
Which is cheaper to run?
Neither model has a hosted API on OpenRouter, so you would run both yourself. Both need about 8 GB of memory at 4-bit quantization.
Which has the longer context window?
Apertus 1.5 8B has the longer context window: 262K tokens against 64K.
Can I use Step3-VL-10B and Apertus 1.5 8B commercially?
Both are released under the Apache-2.0 license. Apache 2.0 and MIT allow commercial use; other licenses can add conditions, so read the license before shipping a product.