# LFM3-1.7B ![Organization](https://img.shields.io/badge/Org-Liquid%20AI-6f42c1) ![Context](https://img.shields.io/badge/Context-256K%20tokens-1f6feb) ![License](https://img.shields.io/badge/License-lfm1.0-2da44e) ![Modality](https://img.shields.io/badge/Modality-Text-8957e5) ![Released](https://img.shields.io/badge/Released-Oct%202026-555) > **Quick answer:** LFM3-1.7B is Liquid AI's 1.73B-parameter text model built on a new "tri-hybrid" architecture (gated short convolutions + sliding-window attention + global attention), released October 2026 with a 256K-token context window. Post-trained for agentic harnesses (Hermes Agent, Pi, OpenClaw), it scores 71.4% on BFCL V4, 80.4% on ToolSandbox, and 96.7% on IFEval — Liquid reports it can power agentic applications that previously needed ~9B-parameter models. At 4-bit precision it needs about 1GB of memory, fitting on a phone or laptop. ## At a Glance **Where LFM3-1.7B leads** - 71.4% BFCL V4 and 80.4% ToolSandbox — ahead of every compared model up to 9B parameters (LFM2.5-2.6B, MiniCPM5-2B, Gemma 4 E4B, Qwen3.5-4B, and even Qwen3.5-9B on BFCL V4) - 96.7% IFEval, 73.0% IFBench, 89.1% Multi-IF — leads all compared models on instruction-following - 45.6 LiveClaw and 41.3 AutomationBench (public) — large agentic-task gains over its LFM2.5-2.6B predecessor (14.7 / 0.9) - ~1GB memory at 4-bit quantization; 2.5x+ faster prefill and 1.8x+ faster decode than compared models across AMD/Apple/iPhone hardware tested **Where it lags** - 63.4% GPQA Diamond — trails Qwen3.5-4B (75.8%) and Qwen3.5-9B (80.4%) on graduate-level science knowledge - 7.2% HLE (no tools) — behind Qwen3.5-9B (16.4%); not recommended by Liquid AI for knowledge-heavy tasks - Not recommended by Liquid AI for agentic coding or complex multi-step reasoning given its small size **Best for:** local/edge agent deployments — tool calling, structured outputs, and domain-specific fine-tuning — where on-device privacy and near-zero marginal inference cost matter more than frontier coding or deep knowledge recall. ## What LFM3-1.7B Is LFM3-1.7B is Liquid AI's newest small model for local agents, introducing a "tri-hybrid" architecture: 24 layers combining 13 gated short-convolution blocks, 5 sliding-window attention blocks, and 6 global attention blocks. It was pre-trained on 22T tokens across two stages (a general mix plus a mid-training phase adding scientific papers, repository-level code, math, and agentic/tool-call traces while extending context to 256K tokens), then post-trained through supervised fine-tuning, mixed reinforcement learning, multi-domain on-policy distillation from specialist RL teachers, and a final agentic-RL stage run inside real harnesses. Liquid positions LFM3-1.7B explicitly as a successor that "can power agentic applications that used to need 9B parameters" — its own benchmark table shows it beating same-class and larger models (Qwen3.5-4B, Qwen3.5-9B) on most agentic, tool-use, and instruction-following evaluations while trailing on knowledge-heavy and long-context-retrieval tasks (GPQA Diamond, HLE, RULER). The model also supports "deferred tools" (tool search), letting a harness keep large tool catalogs out of the prompt until needed — a capability aimed squarely at production agent deployments with many available tools. ## Specifications | Field | Value | |---|---| | Organization | Liquid AI | | Parameters | 1.73B (dense) | | Context window | 262,144 tokens (256K) | | Architecture | Tri-hybrid — 24 layers: 13 gated short-convolution, 5 sliding-window attention, 6 global attention | | License | LFM Open License v1.0 (lfm1.0) | | Release date | October 2026 | | Modality | Text (16 languages) | ## Pricing Open weights, self-hostable — no per-token API pricing. Available on [Hugging Face](https://huggingface.co/LiquidAI/LFM3-1.7B) in BF16, FP8, NVFP4, GGUF (llama.cpp/Ollama/LM Studio), MLX (Apple Silicon), and ONNX formats. At 4-bit quantization, the model uses about 1GB of memory; at 1,024 concurrent requests it can produce 3.6 billion tokens/day on a single NVIDIA B300. ## Public Benchmark Scores | Benchmark | Score | Source | Date | |---|---|---|---| | [BFCL V4](/benchmarks/patil-et-al-uc-berkeley/bfcl-v4) | 71.39 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) | 2026-10 | | [ToolSandbox](/benchmarks/lu-et-al-apple/toolsandbox) | 80.36 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) | 2026-10 | | [MCP-Atlas](/benchmarks/mcp-atlas) | 32.57 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) | 2026-10 | | [AutomationBench](/benchmarks/automationbench) | 41.29 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) (public subset) | 2026-10 | | [Claw-Eval](/benchmarks/cross-lab-agent-evaluation-reference-cited-by-meta-kimi-qwen-tencent/claw-eval) average (EN) | 68.64 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) | 2026-10 | | [SWE-bench Verified](/benchmarks/swe-bench-verified) | 43.40 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) (OpenCode harness) | 2026-10 | | [SWE-bench Multilingual](/benchmarks/swe-bench-multilingual) | 23.00 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) (OpenCode harness) | 2026-10 | | [Terminal-Bench 2.1](/benchmarks/tbench/terminalbench-21) | 12.92 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) (OpenCode harness) | 2026-10 | | [LiveCodeBench v6](/benchmarks/livecodebench/livecodebench-v6) | 73.46 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) | 2026-10 | | [LiveCodeBench Pro](/benchmarks/livecodebench-team/livecodebench-pro) | 31.74 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) (25Q2 subset) | 2026-10 | | [SciCode](/benchmarks/scicode) | 28.47 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) | 2026-10 | | [IFEval](/benchmarks/ifeval) | 96.66 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) | 2026-10 | | [IFBench](/benchmarks/allen-institute-for-ai/ifbench) | 73.03 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) | 2026-10 | | [Multi-IF](/benchmarks/he-et-al-meta/multi-if) | 89.10 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) | 2026-10 | | [GPQA Diamond](/benchmarks/rein-et-al/gpqa-diamond) | 63.37 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) | 2026-10 | | [Humanity's Last Exam](/benchmarks/center-for-ai-safety-scale-ai/humanitys-last-exam) (no tools) | 7.17 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) | 2026-10 | | [AA-Omniscience](/benchmarks/artificial-analysis/aa-omniscience) (public) | 0.50 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) | 2026-10 | | [AA-LCR](/benchmarks/artificial-analysis/aa-lcr) | 52.83 | [LFM3-1.7B model card](https://huggingface.co/LiquidAI/LFM3-1.7B) | 2026-10 | *Scores above are reported by Liquid AI and shown for context. They are not Benchgen measurements. Figures depend on harness, tools, and effort settings — all models in Liquid's comparison table were served with vLLM in BF16 at full context length; see the source for full methodology.* **Excluded from the table above (bucket-C, documented per skill procedure):** - **LiveClaw** (45.60) — Liquid's own OpenClaw-harness agent eval; no independently discoverable public benchmark definition/paper found (same exclusion pattern as LFM2.5-2.6B's prior release). - **BrowseComp-Plus (OpenClaw subset)** (37.00) — a harness-specific subset of BrowseComp-Plus; our local `browsecomp*` entities don't document this exact subset's methodology, so left unmapped rather than guessed. - **WebArena-Verified** (38.42) and **EnterpriseOps-Gym** (28.20) — no local Benchgen benchmark entity exists for either; insufficient public methodology/paper info in the model card alone to stand up new configs this run. - **τ³-bench (4 domains), average** (68.01) — ambiguous generic name vs. our existing domain-specific `tau3-banking` entity; per standing policy, never silently narrow/widen scope. Excluded, same as the Nanbeige4.3-33B precedent. - **IFStruct** (93.72) — Liquid's own structured-instruction-following eval; not an independently verifiable standalone public benchmark (same exclusion as LFM2.5-2.6B). - **Humanity's Last Exam (with tools)** (13.90) and **AA-Omniscience accuracy/non-hallucination sub-metrics** (7.07 / 92.93) — secondary/derived metrics alongside the primary HLE and AA-Omniscience scores already listed above; not separate benchmark entities. - **RULER 128K** (65.56) — no local Benchgen entity for RULER exists yet; skipped rather than creating a new config this run (candidate for a future bucket-B review given it's a widely cross-model-comparable long-context benchmark). ## LFM3-1.7B vs Alternatives | Model | Params | BFCL V4 | ToolSandbox | IFEval | |---|---|---|---|---| | **LFM3-1.7B** | 1.73B | 71.4% | 80.4% | 96.7% | | [LFM2.5-2.6B](/models/liquid-ai/lfm2-5-2-6b) | 2.6B | 55.9% | 78.0% | 95.4% | | [Qwen3.5-4B](/models/alibaba/qwen3-5-4b) | 4.7B | 55.1% | 75.4% | 92.5% | | [Qwen3.5-9B](/models/alibaba/qwen3-5-9b) | 9.7B | 62.8% | 76.4% | 92.1% | Despite being Liquid AI's smallest LFM3-generation model, LFM3-1.7B leads every compared model — including the 9.7B Qwen3.5-9B — on BFCL V4, ToolSandbox, and IFEval, while remaining small enough to run locally on a phone or laptop; the clear trade-off for builders is weaker raw knowledge recall (GPQA Diamond, HLE) versus the larger Qwen3.5 models. ## How LFM3-1.7B Performs on Real Agent Tasks LFM3-1.7B's benchmark spread is heavily weighted toward tool calling, instruction following, and short-horizon agentic tasks (BFCL V4, ToolSandbox, MCP-Atlas, Claw-Eval) rather than deep knowledge or long-horizon coding — consistent with Liquid's explicit guidance that the model is "not recommended for agentic coding or complex multi-step reasoning due to its small size." A high public score on tool-use benchmarks indicates strong general tool-calling reliability, but builders deploying it for a specific production agent (e.g. a fixed set of internal APIs) should still validate accuracy on their own tool schemas and task distribution rather than relying on the aggregate benchmark numbers alone. ## Use LFM3-1.7B via API ```python from transformers import pipeline generator = pipeline( "text-generation", model="LiquidAI/LFM3-1.7B", trust_remote_code=True, ) messages = [{"role": "user", "content": "What is C. elegans?"}] output = generator( messages, do_sample=True, temperature=0.1, top_k=50, repetition_penalty=1.1, max_new_tokens=2048, ) print(output[0]["generated_text"][-1]["content"]) ``` ## Related Models & Benchmarks - Compare: [LFM2.5-2.6B](/models/liquid-ai/lfm2-5-2-6b) · [Qwen3.5-9B](/models/alibaba/qwen3-5-9b) - Benchmarks: [BFCL V4](/benchmarks/patil-et-al-uc-berkeley/bfcl-v4) · [ToolSandbox](/benchmarks/lu-et-al-apple/toolsandbox) · [GPQA Diamond](/benchmarks/rein-et-al/gpqa-diamond) ## Frequently Asked Questions

What is LFM3-1.7B?

LFM3-1.7B is Liquid AI's 1.73B-parameter text model built on a new tri-hybrid architecture (gated short convolutions, sliding-window attention, and global attention), designed for local agents, tool calling, and domain-specific fine-tuning.

What is LFM3-1.7B's context window?

262,144 tokens (256K) for combined input and output — roughly 200,000 words of text in a single request.

How much does LFM3-1.7B cost?

It's released under an open-weight license — free to download; you pay only your own hosting cost. At 4-bit precision it runs on consumer hardware in about 1GB of memory.

Is LFM3-1.7B open source?

Yes — released under Liquid AI's LFM Open License v1.0 (lfm1.0).

What is LFM3-1.7B's knowledge cutoff?

Not disclosed by Liquid AI at release.

--- *Specs and scores sourced from Liquid AI's official Hugging Face model card and release blog post (liquid.ai/blog/lfm3); third-party/comparison-model scores (LFM2.5-2.6B, MiniCPM5-2B, Gemma 4 E4B, Qwen3.5-4B/9B) as reported in the same launch benchmark table. Last updated 2026-10-29.*

LFM3-1.7B — Evaluations

Evaluations