Compare / gpt-oss-20b vs Tri-21B Think
gpt-oss-20b vs Tri-21B Think
Updated November 2026
Verdict
gpt-oss-20b scores higher on the Small LM Index: 64.0 vs 43.6. On the 6 benchmarks both labs report, gpt-oss-20b wins 6 and Tri-21B Think wins 0. Both need about 15.1 GB of memory at 4-bit quantization. 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 mostly write code
- you need math and reasoning
- you need precise instruction following
- 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 Tri-21B Think if…
No clear advantage in the data we have.
Specifications
| gpt-oss-20b | Tri-21B Think | |
|---|---|---|
| Small LM Index | 64.0 (likely 60–67) | 43.6 (likely 40–48) |
| Lab | OpenAI · United States | Trillion Labs · Rest of Asia |
| Released | 2025-08-05 | 2026-02-18 |
| Parameters | 20.9B (3.6B active, MoE) | 20.7B (dense) |
| Memory at 4-bit | ~13.8 GB | ~15.1 GB |
| Context window | 131K tokens | 32K tokens |
| Input | text | text |
| Reasoning | Always thinks | Always thinks |
| 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 | 459 |
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 | Tri-21B Think |
|---|---|---|
| MMLU-Pro Knowledge | 75.0* | 74.3 |
| GPQA Diamond Knowledge | 71.5 | 62.6 |
| Humanity's Last Exam Reasoning | 10.9 | 5.5 |
| AIME Math | 91.7 | 56.7 |
| LiveCodeBench Coding | 61.0* | 53.7 |
| SWE-bench Verified Coding | 60.7 | ≈ 24 |
| IFBench Instruction following | 65.0* | 56.7 |
| IFEval Instruction following | — | 84.7 |
| 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, gpt-oss-20b or Tri-21B Think?
gpt-oss-20b scores higher on the Small LM Index: 64.0 vs 43.6. On the 6 benchmarks both labs report, gpt-oss-20b wins 6 and Tri-21B Think 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. Both need about 15.1 GB of memory at 4-bit quantization.
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 Tri-21B Think 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.