Compare / Spark-X2.5 4B vs Qwen3.5 4B
Spark-X2.5 4B vs Qwen3.5 4B
Updated November 2026
Verdict
Spark-X2.5 4B scores higher on the Small LM Index: 65.5 vs 57.8, although their likely ranges overlap, so the lead is not certain. On the 3 benchmarks both labs report, Spark-X2.5 4B wins 1 and Qwen3.5 4B wins 2. Both need about 3.4 GB of memory at 4-bit quantization. Neither model has a hosted API on OpenRouter, so you would run both yourself.
Pick Spark-X2.5 4B if…
- you want the strongest model overall
- you need precise instruction following
- you work with long documents
Pick Qwen3.5 4B if…
- you build agents that call tools
- you need image input
- you want the more widely used model (more quantized builds and community support)
Specifications
| Spark-X2.5 4B | Qwen3.5 4B | |
|---|---|---|
| Small LM Index | 65.5 (likely 63–68) | 57.8 (likely 52–63) |
| Lab | SparkLLM (XHToken) · China | Alibaba (Qwen) · China |
| Released | 2026-08-24 | 2026-03-02 |
| Parameters | 4.1B (dense) | 4B (dense) |
| Memory at 4-bit | ~3.4 GB | ~3.3 GB |
| Context window | 1M tokens | 262K tokens |
| Input | text | text, image, video |
| Reasoning | Hybrid (thinking on/off) | 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) | 30.3K | 6.1M |
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 | Spark-X2.5 4B | Qwen3.5 4B |
|---|---|---|
| MMLU-Pro Knowledge | ≈ 82 | 79.1 |
| GPQA Diamond Knowledge | 73.7 | 76.2 |
| Humanity's Last Exam Reasoning | 12.3 | ≈ 9 |
| AIME Math | 90.7 | ≈ 81 |
| LiveCodeBench Coding | ≈ 75 | 55.8 |
| SWE-bench Verified Coding | 41.6 | ≈ 42 |
| τ²-Bench Agentic | 75.1 | 79.9 |
| IFBench Instruction following | 75.0 | 59.2 |
| MMMU-Pro Vision | — | 65.4✓ |
| IFEval Instruction following | 93.0 | — |
| BFCL Agentic | 65.1 | — |
| HMMT Math | 81.2 | — |
| AA-LCR Long context | 56.3 | — |
| SciCode Coding | 34.7 | — |
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, Spark-X2.5 4B or Qwen3.5 4B?
Spark-X2.5 4B scores higher on the Small LM Index: 65.5 vs 57.8, although their likely ranges overlap, so the lead is not certain. On the 3 benchmarks both labs report, Spark-X2.5 4B wins 1 and Qwen3.5 4B wins 2.
Which is cheaper to run?
Neither model has a hosted API on OpenRouter, so you would run both yourself. Both need about 3.4 GB of memory at 4-bit quantization.
Which has the longer context window?
Spark-X2.5 4B has the longer context window: 1M tokens against 262K.
Can I use Spark-X2.5 4B and Qwen3.5 4B 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.
Related comparisons
- Nanbeige4.2 3B vs Spark-X2.5 4B
- Spark-X2.5 4B vs Spark-X2.6 4B
- Qwen3 4B Thinking 2507 vs Spark-X2.5 4B
- Nemotron Labs Dory 1.0 4B vs Spark-X2.5 4B
- Gemma 3 4B vs Spark-X2.5 4B
- Nanbeige4.2 3B vs Qwen3.5 4B
- Qwen3.5 4B vs Spark-X2.6 4B
- Nemotron Labs Dory 1.0 4B vs Qwen3.5 4B
- Qwen3 4B Thinking 2507 vs Qwen3.5 4B
- Gemma 3 4B vs Qwen3.5 4B