Model
LFM2.5-1.2B-Base
lfm2-5-1-2b-baseRecord
- Family
- lfm
- Architecture
- dense
- Parameters
- 1.17
- Active parameters
- 1.17
- HF downloads / 30d
- 391,220
Hugging Face model card
Identity
https://huggingface.co/LiquidAI/LFM2.5-1.2B-Instruct- Repository
- LiquidAI/LFM2.5-1.2B-Instruct
- Status
- known
- Link type
- Exact Hub repository
Public Hugging Face repository confirmed by the Hub API.
Leaderboard scores
Scraped public quality scores joined to this model by its Hugging Face repository. They describe the listed variant, not any specific local quantization, and are never speed measurements.
| Benchmark | Category | Score | Rank | Scored variant | Confidence |
|---|---|---|---|---|---|
| if-eval | instruction-following | 86.23 | 66 | LFM2.5-1.2B-Instruct-GGUF | low |
| if-eval | instruction-following | 86.23 | 67 | LFM2.5-1.2B-Instruct | low |
| if-eval | instruction-following | 86.23 | 69 | LFM2.5-1.2B-Instruct-GGUF | low |
| if-eval | instruction-following | 86.23 | 70 | LFM2.5-1.2B-Thinking-GGUF | low |
| if-eval | instruction-following | 52.29 | 266 | LFM2.5-VL-1.6B-GGUF | low |
| ifbench | instruction-following | 47.33 | 46 | LFM2.5-1.2B-Instruct-GGUF | low |
| ifbench | instruction-following | 47.33 | 47 | LFM2.5-1.2B-Instruct | low |
| ifbench | instruction-following | 47.33 | 49 | LFM2.5-1.2B-Instruct-GGUF | low |
| ifbench | instruction-following | 47.33 | 50 | LFM2.5-1.2B-Thinking-GGUF | low |
| mmlu-pro | knowledge | 44.35 | 374 | LFM2.5-1.2B-Instruct-GGUF | low |
| mmlu-pro | knowledge | 44.35 | 375 | LFM2.5-1.2B-Instruct | low |
| mmlu-pro | knowledge | 44.35 | 377 | LFM2.5-1.2B-Instruct-GGUF | low |
| mmlu-pro | knowledge | 44.35 | 378 | LFM2.5-1.2B-Thinking-GGUF | low |
| aime-2025 | math | 14 | 245 | LFM2.5-1.2B-Instruct-GGUF | low |
| aime-2025 | math | 14 | 246 | LFM2.5-1.2B-Instruct | low |
| aime-2025 | math | 14 | 248 | LFM2.5-1.2B-Instruct-GGUF | low |
| aime-2025 | math | 14 | 249 | LFM2.5-1.2B-Thinking-GGUF | low |
| gsm8k | math | 64.52 | 216 | LFM2.5-1.2B-Thinking-GGUF | low |
| math-500 | math | 63.2 | 130 | LFM2.5-1.2B-Thinking-GGUF | low |
| mmbench | multimodal | 65.9 | 60 | LFM2.5-VL-1.6B-GGUF | low |
| mmmu | multimodal | 40.56 | 199 | LFM2.5-VL-1.6B-GGUF | low |
| gpqa | reasoning | 38.89 | 183 | LFM2.5-1.2B-Instruct-GGUF | low |
| gpqa | reasoning | 38.89 | 184 | LFM2.5-1.2B-Instruct | low |
| gpqa | reasoning | 38.89 | 185 | LFM2.5-1.2B-Instruct-GGUF | low |
| gpqa-diamond | reasoning | 38.89 | 301 | LFM2.5-1.2B-Thinking-GGUF | low |
Remaining fields
Identity, launch, related records, and measured speed are shown above. This is the rest of the normalized record.
- active params
- 1.17
- architecture
- dense
- family
- lfm
- id
- lfm2-5-1-2b-base
- name
- LFM2.5-1.2B-Base
- params
- 1.17
- schema version
- local-ai-registry/v1
downloads
- all time
- 2,486,870
- captured at
- 2026-08-31T15:38:41Z
- last 30d
- 391,220
Provenance & metadata (1)
source huggingface-api
Provenance & metadata (2)
facts
- active params · provenance · captured at
- 2026-09-01T00:18:17Z
active params · reason all-parameters-active-in-dense-model
active params · state known
- architecture · provenance · captured at
- 2026-08-31T23:49:15Z
architecture · reason resolved-huggingface-config
architecture · state known
- family · provenance · captured at
- 2026-08-31T23:43:51Z
family · reason canonical-model-name-family
family · state known
- params · provenance · captured at
- 2026-09-01T00:18:17Z
params · reason safetensors-parameter-metadata
params · state known
provenance
- captured at
- 2026-08-30T09:10:02Z
sources
| captured at | kind | url |
|---|---|---|
| 2026-08-30T09:10:02Z | normalized-model | huggingface.co/LiquidAI/LFM2.5-1.2B-Instruct ↗ |
| 2026-09-01T00:18:17Z | huggingface-api | huggingface.co/api/models/LiquidAI/LFM2.5-1.2B-Instruct ↗ |