Compare / Edge0 35B-A3B (preview) vs Seed-OSS 36B Instruct
Edge0 35B-A3B (preview) vs Seed-OSS 36B Instruct
Updated November 2026
Verdict
Edge0 35B-A3B (preview) scores higher on the Small LM Index: 62.8 vs 60.3, although their likely ranges overlap, so the lead is not certain. On the 3 benchmarks both labs report, Edge0 35B-A3B (preview) wins 2 and Seed-OSS 36B Instruct wins 1. Edge0 35B-A3B (preview) is smaller: it needs about 22.4 GB of memory at 4-bit, against 25.3 GB for Seed-OSS 36B Instruct. Neither model has a hosted API on OpenRouter, so you would run both yourself.
Pick Edge0 35B-A3B (preview) if…
- you want the strongest model overall
- you need math and reasoning
- you have limited memory (about 22.4 GB at 4-bit)
- you need image input
Pick Seed-OSS 36B Instruct if…
- you work with long documents
- you want the more widely used model (more quantized builds and community support)
Specifications
| Edge0 35B-A3B (preview) | Seed-OSS 36B Instruct | |
|---|---|---|
| Small LM Index | 62.8 (likely 59–67) | 60.3 (likely 57–64) |
| Lab | Edge0 · Other | ByteDance Seed · China |
| Released | 2026-09-08 | 2025-08-20 |
| Parameters | 34.7B (3B active, MoE) | 36.2B (dense) |
| Memory at 4-bit | ~22.4 GB | ~25.3 GB |
| Context window | 262K tokens | 512K tokens |
| Input | text, image | text |
| Reasoning | Hybrid (thinking on/off) | Always thinks |
| License | Apache-2.0 | Apache-2.0 |
| API price / 1M tokens | Self-host only | Self-host only |
| Hugging Face downloads (30 days) | 4.2K | 40.7K |
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 | Edge0 35B-A3B (preview) | Seed-OSS 36B Instruct |
|---|---|---|
| MMLU-Pro Knowledge | 81.0 | 82.7 |
| GPQA Diamond Knowledge | 79.8 | 71.4 |
| Humanity's Last Exam Reasoning | ≈ 12 | 10.1 |
| AIME Math | 86.6 | 84.7 |
| LiveCodeBench Coding | ≈ 72 | 67.4 |
| SWE-bench Verified Coding | ≈ 50 | 56.0 |
| IFBench Instruction following | 57.9 | — |
| MMLU Knowledge | — | 87.4 |
| Multilingual MMLU Multilingual | — | 78.4 |
| IFEval Instruction following | — | 85.8 |
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, Edge0 35B-A3B (preview) or Seed-OSS 36B Instruct?
Edge0 35B-A3B (preview) scores higher on the Small LM Index: 62.8 vs 60.3, although their likely ranges overlap, so the lead is not certain. On the 3 benchmarks both labs report, Edge0 35B-A3B (preview) wins 2 and Seed-OSS 36B Instruct wins 1.
Which is cheaper to run?
Neither model has a hosted API on OpenRouter, so you would run both yourself. Edge0 35B-A3B (preview) is smaller: it needs about 22.4 GB of memory at 4-bit, against 25.3 GB for Seed-OSS 36B Instruct.
Which has the longer context window?
Seed-OSS 36B Instruct has the longer context window: 512K tokens against 262K.
Can I use Edge0 35B-A3B (preview) and Seed-OSS 36B Instruct 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.