24B
- input
- $0.08
- cached
- $0.02
- output
- $0.20
- Context
- 262Ktokens
- Listed
- 6d ago
- Size
- 1.7Bparameters
Cost/1M tokens
LFM3-1.7B is Liquid AI's post-trained text model in the new LFM3 generation of its Liquid Foundation Models, fine-tuned from LFM3-1.7B-Base. The launch announcement bills LFM3 as a multimodal family for specialized intelligence, claiming its redesigned hybrid architecture is far more expressive than LFM2 while staying efficient and about 25% faster than LFM2-1.2B at 8K tokens of context. It targets local agents, tool calling, and domain-specific edge fine-tuning, and the model card advises against agentic coding and complex multi-step reasoning given its size.
- Language model
- Open weights · 1.7B
- Reasoning
- Tools
- Structured output
How it compares
Size class
Tiny1.7BAn open-weight model under 5 billion parameters, small enough for phones, laptops and edge devices.
Open models of a similar size from other labs
Vs. your stack
Add the models you use to see how this one compares.
Sign inFrequently asked questions
Running locally
Estimated VRAM is 1.7 GB at Q4, 2.6 GB at Q8 and 4.4 GB at FP16, derived from the 1.7B parameter count. All figures are estimates, covering weights plus about 15% and 0.5 GB for a short context.
Yes. A 24 GB, 16 GB, 12 GB or 8 GB GPU all fit the model at FP16, with an estimated 4.4 GB of VRAM.
Yes. Estimated VRAM at Q4 is 1.7 GB, and assuming macOS lets the GPU use about 75% of unified memory, the smallest common Mac memory size that fits Q4 is 8 GB (estimate).
No support is recorded for llama.cpp, vLLM, LM Studio or MLX. Ollama is the only runtime with recorded support, and no Ollama builds are published.
API and pricing
Input costs $0.08 per 1M tokens and output costs $0.20 per 1M tokens, based on 1 paid provider tracked.
Yes, cached input costs $0.02 per 1M tokens versus $0.08 per 1M tokens uncached, a 75% discount.
A request with 10,000 input tokens and 1,000 output tokens costs $0.001 at the cheapest provider, Liquid.
Of the 1 endpoint with uptime data in the last day, 1 was at 99% uptime or better. The best uptime was 100.0%, by Liquid.
Privacy and data
Liquid is the provider with zero data retention (ZDR eligible or 0-day retention).
Yes, LFM3-1.7B has open weights, so you can self-host it for full control of your data.
Capabilities and specs
The context window is 262K tokens (262,144) and the maximum output is 8K tokens (8,192).
No. The only input modality is text.
Yes, tool calling (function calling) is supported.
Yes, structured output (JSON schema or JSON mode) is supported.
Yes, a reasoning mode is supported and reasoning is always on. 1 of 1 tracked providers expose reasoning, and no effort setting (reasoning_effort) is offered by a provider.
Quality and benchmarks
Matched on model kind, blended price and release date: Ternary Bonsai 2 9B by PrismML at $0.0925 per 1M tokens, Nex-N2.5-Pro by Nex AGI at $0.1188 per 1M tokens and Nemotron 3.5 Lightning by NVIDIA at $0.1025 per 1M tokens.
Ternary Bonsai 2 9BLFM3-1.7B vs Ternary Bonsai 2 9BNex-N2.5-ProLFM3-1.7B vs Nex-N2.5-ProNemotron 3.5 LightningLFM3-1.7B vs Nemotron 3.5 Lightning
Licensing and access
Open weights.
Download from the Hugging Face repository LiquidAI/LFM3-1.7B.
Basics
LFM3-1.7B was made by Liquid AI and was first listed on OpenRouter on November 6, 2026.
No, the listed API price is $0.08 input and $0.20 output per 1M tokens, though the model has open weights.
Generated from Vertical Tabs data, updated November 12, 2026.
Discussion
No comments yet
Be the first to share your thoughts.