Compare / Tri-21B Think vs EuroLLM 22B Instruct
Tri-21B Think vs EuroLLM 22B Instruct
Updated November 2026
Verdict
Tri-21B Think scores higher on the Small LM Index: 43.6 vs 19.2. On the 3 benchmarks both labs report, Tri-21B Think wins 3 and EuroLLM 22B Instruct wins 0. Both need about 16.5 GB of memory at 4-bit quantization. Neither model has a hosted API on OpenRouter, so you would run both yourself.
Pick Tri-21B Think if…
- you want the strongest model overall
- you need precise instruction following
Pick EuroLLM 22B Instruct if…
No clear advantage in the data we have.
Specifications
| Tri-21B Think | EuroLLM 22B Instruct | |
|---|---|---|
| Small LM Index | 43.6 (likely 40–48) | 19.2 (likely 16–24) |
| Lab | Trillion Labs · Rest of Asia | EuroLLM (UTTER consortium) · Europe |
| Released | 2026-02-18 | 2025-12-05 |
| Parameters | 20.7B (dense) | 22.6B (dense) |
| Memory at 4-bit | ~15.1 GB | ~16.5 GB |
| Context window | 32K tokens | 32K tokens |
| Input | text | text |
| Reasoning | Always thinks | No thinking |
| License | Apache-2.0 | Apache-2.0 |
| API price / 1M tokens | Self-host only | Self-host only |
| Hugging Face downloads (30 days) | 459 | 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 | Tri-21B Think | EuroLLM 22B Instruct |
|---|---|---|
| MMLU-Pro Knowledge | 74.3 | 50.9 |
| GPQA Diamond Knowledge | 62.6 | 26.8 |
| Humanity's Last Exam Reasoning | 5.5 | ≈ 1 |
| AIME Math | 56.7 | ≈ 15 |
| LiveCodeBench Coding | 53.7 | ≈ 19 |
| IFBench Instruction following | 56.7 | — |
| MMLU Knowledge | — | 69.8 |
| Multilingual MMLU Multilingual | — | 64.1 |
| IFEval Instruction following | 84.7 | 67.2 |
| MATH-500 Math | — | 54.5 |
| AA-LCR Long context | 11.0 | — |
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, Tri-21B Think or EuroLLM 22B Instruct?
Tri-21B Think scores higher on the Small LM Index: 43.6 vs 19.2. On the 3 benchmarks both labs report, Tri-21B Think wins 3 and EuroLLM 22B Instruct wins 0.
Which is cheaper to run?
Neither model has a hosted API on OpenRouter, so you would run both yourself. Both need about 16.5 GB of memory at 4-bit quantization.
Which has the longer context window?
Both have a 32K-token context window.
Can I use Tri-21B Think 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.