Compare / gpt-oss-20b vs EuroLLM 22B Instruct
gpt-oss-20b vs EuroLLM 22B Instruct
Updated November 2026
Verdict
gpt-oss-20b scores higher on the Small LM Index: 64.0 vs 19.2. On the 2 benchmarks both labs report, gpt-oss-20b wins 2 and EuroLLM 22B Instruct wins 0. gpt-oss-20b is smaller: it needs about 13.8 GB of memory at 4-bit, against 16.5 GB for EuroLLM 22B Instruct. Only gpt-oss-20b has a hosted API on OpenRouter ($0.020 per million tokens); the other has to be self-hosted.
Pick gpt-oss-20b if…
- you want the strongest model overall
- you have limited memory (about 13.8 GB at 4-bit)
- you want a hosted API rather than running it yourself
- you work with long documents
- you want the more widely used model (more quantized builds and community support)
Pick EuroLLM 22B Instruct if…
No clear advantage in the data we have.
Specifications
| gpt-oss-20b | EuroLLM 22B Instruct | |
|---|---|---|
| Small LM Index | 64.0 (likely 60–67) | 19.2 (likely 16–24) |
| Lab | OpenAI · United States | EuroLLM (UTTER consortium) · Europe |
| Released | 2025-08-05 | 2025-12-05 |
| Parameters | 20.9B (3.6B active, MoE) | 22.6B (dense) |
| Memory at 4-bit | ~13.8 GB | ~16.5 GB |
| Context window | 131K tokens | 32K tokens |
| Input | text | text |
| Reasoning | Always thinks | No thinking |
| License | Apache-2.0 | Apache-2.0 |
| API price / 1M tokens | $0.020 ($0.011 in, $0.045 out) | Self-host only |
| Hugging Face downloads (30 days) | 6M | 8.5K |
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 | gpt-oss-20b | EuroLLM 22B Instruct |
|---|---|---|
| MMLU-Pro Knowledge | 75.0* | 50.9 |
| GPQA Diamond Knowledge | 71.5 | 26.8 |
| Humanity's Last Exam Reasoning | 10.9 | ≈ 1 |
| AIME Math | 91.7 | ≈ 15 |
| LiveCodeBench Coding | 61.0* | ≈ 19 |
| SWE-bench Verified Coding | 60.7 | ≈ 5 |
| IFBench Instruction following | 65.0* | — |
| MMLU Knowledge | — | 69.8 |
| Multilingual MMLU Multilingual | — | 64.1 |
| IFEval Instruction following | — | 67.2 |
| MATH-500 Math | — | 54.5 |
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, gpt-oss-20b or EuroLLM 22B Instruct?
gpt-oss-20b scores higher on the Small LM Index: 64.0 vs 19.2. On the 2 benchmarks both labs report, gpt-oss-20b wins 2 and EuroLLM 22B Instruct wins 0.
Which is cheaper to run?
Only gpt-oss-20b has a hosted API on OpenRouter ($0.020 per million tokens); the other has to be self-hosted. gpt-oss-20b is smaller: it needs about 13.8 GB of memory at 4-bit, against 16.5 GB for EuroLLM 22B Instruct.
Which has the longer context window?
gpt-oss-20b has the longer context window: 131K tokens against 32K.
Can I use gpt-oss-20b and EuroLLM 22B 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.