Compare / Kolibri-1 vs K2-V2 Instruct
Kolibri-1 vs K2-V2 Instruct
Updated November 2026
Verdict
Kolibri-1 scores higher on the Small LM Index: 68.3 vs 57.9. On the 2 benchmarks both labs report, Kolibri-1 wins 2 and K2-V2 Instruct wins 0. K2-V2 Instruct is smaller: it needs about 47 GB of memory at 4-bit, against 49.6 GB for Kolibri-1. Neither model has a hosted API on OpenRouter, so you would run both yourself.
Pick Kolibri-1 if…
- you want the strongest model overall
- you need math and reasoning
- you work with long documents
- you want the more widely used model (more quantized builds and community support)
Pick K2-V2 Instruct if…
- you have limited memory (about 47 GB at 4-bit)
Specifications
| Kolibri-1 | K2-V2 Instruct | |
|---|---|---|
| Small LM Index | 68.3 (likely 67–70) | 57.9 (likely 54–62) |
| Lab | Aleph Alpha · Europe | LLM360 / MBZUAI (K2) · Other |
| Released | 2026-10-03 | 2025-12-03 |
| Parameters | 78.1B (3.5B active, MoE) | 70B (dense) |
| Memory at 4-bit | ~49.6 GB | ~47 GB |
| Context window | 1M tokens | 512K tokens |
| Input | text | text |
| Reasoning | Always thinks | 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) | 22.8K | 1.2K |
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 | Kolibri-1 | K2-V2 Instruct |
|---|---|---|
| MMLU-Pro Knowledge | 80.0 | ≈ 77 |
| GPQA Diamond Knowledge | 84.3 | 69.3 |
| Humanity's Last Exam Reasoning | 21.5 | ≈ 9 |
| AIME Math | 96.0 | 80.2 |
| LiveCodeBench Coding | 85.9 | ≈ 67 |
| SWE-bench Verified Coding | 66.4 | ≈ 42 |
| Terminal-Bench Agentic | 27.7 | — |
| τ²-Bench Agentic | 80.4 | — |
| IFBench Instruction following | 78.1 | — |
| IFEval Instruction following | — | 90.0 |
| BFCL Agentic | 61.4 | — |
| HMMT Math | — | 71.4 |
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, Kolibri-1 or K2-V2 Instruct?
Kolibri-1 scores higher on the Small LM Index: 68.3 vs 57.9. On the 2 benchmarks both labs report, Kolibri-1 wins 2 and K2-V2 Instruct wins 0.
Which is cheaper to run?
Neither model has a hosted API on OpenRouter, so you would run both yourself. K2-V2 Instruct is smaller: it needs about 47 GB of memory at 4-bit, against 49.6 GB for Kolibri-1.
Which has the longer context window?
Kolibri-1 has the longer context window: 1M tokens against 512K.
Can I use Kolibri-1 and K2-V2 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.