Compare / LFM2.5 1.2B Thinking vs LFM2.5 1.2B Instruct
LFM2.5 1.2B Thinking vs LFM2.5 1.2B Instruct
Updated November 2026
Verdict
LFM2.5 1.2B Thinking scores higher on the Small LM Index: 30.1 vs 20.6. On the 6 benchmarks both labs report, LFM2.5 1.2B Thinking wins 4 and LFM2.5 1.2B Instruct wins 2. Both need about 1.3 GB of memory at 4-bit quantization. Neither model has a hosted API on OpenRouter, so you would run both yourself.
Pick LFM2.5 1.2B Thinking if…
- you want the strongest model overall
- you need math and reasoning
- you build agents that call tools
Pick LFM2.5 1.2B Instruct if…
- you want the more widely used model (more quantized builds and community support)
Specifications
| LFM2.5 1.2B Thinking | LFM2.5 1.2B Instruct | |
|---|---|---|
| Small LM Index | 30.1 (likely 27–34) | 20.6 (likely 18–24) |
| Lab | Liquid AI · United States | Liquid AI · United States |
| Released | 2026-01-20 | 2026-01-06 |
| Parameters | 1.2B (dense) | 1.2B (dense) |
| Memory at 4-bit | ~1.3 GB | ~1.3 GB |
| Context window | 32K tokens | 32K tokens |
| Input | text | text |
| Reasoning | Always thinks | No thinking |
| License | LFM Open License | LFM Open License |
| API price / 1M tokens | Self-host only | Self-host only |
| Hugging Face downloads (30 days) | 8.1K | 122.7K |
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 | LFM2.5 1.2B Thinking | LFM2.5 1.2B Instruct |
|---|---|---|
| MMLU-Pro Knowledge | 49.7 | 44.4 |
| GPQA Diamond Knowledge | 37.9 | 38.9 |
| AIME Math | 31.7 | 14.0 |
| IFBench Instruction following | 44.9 | 47.3 |
| IFEval Instruction following | 88.4 | 86.2 |
| BFCL Agentic | 57.0 | 49.1 |
| MATH-500 Math | 88.0 | — |
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, LFM2.5 1.2B Thinking or LFM2.5 1.2B Instruct?
LFM2.5 1.2B Thinking scores higher on the Small LM Index: 30.1 vs 20.6. On the 6 benchmarks both labs report, LFM2.5 1.2B Thinking wins 4 and LFM2.5 1.2B Instruct wins 2.
Which is cheaper to run?
Neither model has a hosted API on OpenRouter, so you would run both yourself. Both need about 1.3 GB of memory at 4-bit quantization.
Which has the longer context window?
Both have a 32K-token context window.
Can I use LFM2.5 1.2B Thinking and LFM2.5 1.2B Instruct commercially?
Both are released under the LFM Open License license. Apache 2.0 and MIT allow commercial use; other licenses can add conditions, so read the license before shipping a product.
Related comparisons
- LFM2.5 1.2B Thinking vs Spark-X2.6 1.7B
- EXAONE 4.0 1.2B vs LFM2.5 1.2B Thinking
- Gemma 3 1B vs LFM2.5 1.2B Thinking
- LFM2.5 1.2B Thinking vs Qwen3.5 0.8B
- LFM2.5 1.2B Thinking vs LFM2.5 350M
- EXAONE 4.0 1.2B vs LFM2.5 1.2B Instruct
- Gemma 3 1B vs LFM2.5 1.2B Instruct
- LFM2.5 1.2B Instruct vs Qwen3.5 0.8B
- LFM2.5 1.2B Instruct vs LFM2.5 350M