# LFM2.5 2.6B

14 Hacker News commenters recommended LFM2.5 2.6B and 8 criticized it, November 2025 to November 2026. It was recommended in 9 of 13 months.

**Verdict:** Fast small model people run locally for classification, security triage, and simple tool use on modest hardware. Critics report coherence problems and say it is weaker than similar-sized rivals.

## Why people recommend it

- Fast on modest or older hardware (3 comments): "I love these smaller models for the added capability they can bring to older hardware." ([source](https://news.ycombinator.com/item?id=49262585)) "they were so fast that you could use them in a similar way" ([source](https://news.ycombinator.com/item?id=49995958)) "With good speed even on a 12th gen Intel." ([source](https://news.ycombinator.com/item?id=49254396))
- Good at classification and security tasks (3 comments): "You can recreate most of those with LFM 2.5 classifier locally" ([source](https://news.ycombinator.com/item?id=49764474)) "you can just prompt for classification" ([source](https://news.ycombinator.com/item?id=50265010)) "handle their security job nicely" ([source](https://news.ycombinator.com/item?id=47457956))
- Vision variant for vision tasks (2 comments): "LiquidAI's LFM 2.5 uses novel techniques to eliminate doom loops, and they have a vision variant" ([source](https://news.ycombinator.com/item?id=48095563)) "My prefer is LFM 450M for vision task, QWEN 9B Q4 for Orchestration" ([source](https://news.ycombinator.com/item?id=47457579))
- Pairs with larger model for orchestration (2 comments): "My prefer is LFM 450M for vision task, QWEN 9B Q4 for Orchestration" ([source](https://news.ycombinator.com/item?id=47457579)) "So I like LFM 450M + QWEN 9B Q4" ([source](https://news.ycombinator.com/item?id=47457758))

## What people warn about

- Serious coherence issues (2 comments): "They have serious issues with coherence." ([source](https://news.ycombinator.com/item?id=49255135)) "suffering from serious coherence issues" ([source](https://news.ycombinator.com/item?id=47984424))
- Not competitive with similar-sized models (2 comments): "I don't think it's even competitive with Gemma 4 E2B" ([source](https://news.ycombinator.com/item?id=49259466)) "It's not even competitive with 2x sized Qwen 4B." ([source](https://news.ycombinator.com/item?id=49254399))
- Often underperforms in practice (2 comments): "These LiquidAI models have never worked well for me in practice." ([source](https://news.ycombinator.com/item?id=49253590)) "2.5 version was too dumb for HA" ([source](https://news.ycombinator.com/item?id=50264200))

## Alternatives hackers recommend

- [Claude Haiku](https://hackerapproved.com/p/claude-haiku.md): 218 recommend
- [OpenAI Luna](https://hackerapproved.com/p/luna.md): 218 recommend
- [Apple Foundation Models](https://hackerapproved.com/p/apple-foundation-models.md): 45 recommend
- [Ministral](https://hackerapproved.com/p/ministral.md): 18 recommend
- [Microsoft Phi](https://hackerapproved.com/p/phi.md): 16 recommend

## Cite as

Hacker Approved, "LFM2.5 2.6B", https://hackerapproved.com/p/lfm2-5-2-6b/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).