Compare / Qwen3.6 35B-A3B vs Hermes 4.3 36B
Qwen3.6 35B-A3B vs Hermes 4.3 36B
Updated November 2026
Verdict
Qwen3.6 35B-A3B scores higher on the Small LM Index: 72.2 vs 51.4. On the 3 benchmarks both labs report, Qwen3.6 35B-A3B wins 3 and Hermes 4.3 36B wins 0. Qwen3.6 35B-A3B is smaller: it needs about 22.6 GB of memory at 4-bit, against 25.3 GB for Hermes 4.3 36B. Only Qwen3.6 35B-A3B has a hosted API on OpenRouter ($0.031 per million tokens); the other has to be self-hosted.
Pick Qwen3.6 35B-A3B if…
- you want the strongest model overall
- you need math and reasoning
- you have limited memory (about 22.6 GB at 4-bit)
- you want a hosted API rather than running it yourself
- you need image input
- you want the more widely used model (more quantized builds and community support)
Pick Hermes 4.3 36B if…
- you work with long documents
Specifications
| Qwen3.6 35B-A3B | Hermes 4.3 36B | |
|---|---|---|
| Small LM Index | 72.2 (likely 71–74) | 51.4 (likely 47–56) |
| Lab | Alibaba (Qwen) · China | Nous Research (Hermes) · United States |
| Released | 2026-04-16 | 2025-11-17 |
| Parameters | 35B (3B active, MoE) | 36.2B (dense) |
| Memory at 4-bit | ~22.6 GB | ~25.3 GB |
| Context window | 262K tokens | 512K tokens |
| Input | text, image, video | text |
| Reasoning | Hybrid (thinking on/off) | Hybrid (thinking on/off) |
| License | Apache-2.0 | Apache-2.0 |
| API price / 1M tokens | $0.031 ($0.010 in, $0.092 out) | Self-host only |
| Hugging Face downloads (30 days) | 4.5M | 2.2K |
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 | Qwen3.6 35B-A3B | Hermes 4.3 36B |
|---|---|---|
| MMLU-Pro Knowledge | 85.2 | 80.7 |
| GPQA Diamond Knowledge | 86.0 | 65.5 |
| Humanity's Last Exam Reasoning | 21.4 | ≈ 7 |
| AIME Math | 92.7 | 69.3 |
| LiveCodeBench Coding | 80.4 | ≈ 59 |
| SWE-bench Verified Coding | 73.4 | ≈ 33 |
| Terminal-Bench Agentic | 51.5 | — |
| MMMU-Pro Vision | 75.3 | — |
| MMLU Knowledge | — | 87.7 |
| IFEval Instruction following | — | 77.9 |
| MATH-500 Math | — | 93.8 |
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, Qwen3.6 35B-A3B or Hermes 4.3 36B?
Qwen3.6 35B-A3B scores higher on the Small LM Index: 72.2 vs 51.4. On the 3 benchmarks both labs report, Qwen3.6 35B-A3B wins 3 and Hermes 4.3 36B wins 0.
Which is cheaper to run?
Only Qwen3.6 35B-A3B has a hosted API on OpenRouter ($0.031 per million tokens); the other has to be self-hosted. Qwen3.6 35B-A3B is smaller: it needs about 22.6 GB of memory at 4-bit, against 25.3 GB for Hermes 4.3 36B.
Which has the longer context window?
Hermes 4.3 36B has the longer context window: 512K tokens against 262K.
Can I use Qwen3.6 35B-A3B and Hermes 4.3 36B 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.