LFM3-1.7B
LFM3-1.7B is Liquid AI's open-weight 1.73B text model for local agents and specialized applications, the text post-trained checkpoint of the LFM3 family (Oct 27, 2026). New tri-hybrid architecture: 24 layers (13 gated short convolutions, 5 sliding-window attention with 256-token window, 6 global attention), head-wise gated attention, partial RoPE, sandwich norm. Pre-trained on 22T tokens; 256K context; 128K vocabulary; 16 languages. Post-trained (SFT, mixed RL, multi-domain on-policy distillation, agentic RL) for harnesses like Hermes Agent, Pi, and OpenClaw; reasoning on by default; supports deferred tools / tool search. Fits in ~1 GB at 4-bit. Vendor says it is not recommended for agentic coding or complex multi-step reasoning. Hugging Face: LiquidAI/LFM3-1.7B (+ Base, FP8, NVFP4, GGUF, MLX, ONNX). License: LFM Open License v1.0. Catalog benches are self-reported from the vendor table. No catalog id (recorded here only): LiveClaw 45.60, Claw-Eval pass^3 51.40, BrowseComp-Plus (OpenClaw subset) 37.00, EnterpriseOps-Gym 28.20, ToolSandbox 80.36, IFStruct 93.72, LiveCodeBench Pro 25Q2 31.74 (pass rate; catalog id is Elo), AA-Omniscience accuracy 7.07 and non-hallucination 92.93. Launch partners include OpenRouter and Vercel, but no gateway id or price was listed at time of adding, so no provider row. Sources: https://www.liquid.ai/blog/lfm3 · https://huggingface.co/LiquidAI/LFM3-1.7B.
Identité
| Organisation | Liquid AI |
|---|---|
| Pays | — |
| Famille | — |
| Type | llm |
| Date de sortie | 2026-10-27 |
| Licence | lfm open license v1.0 |
| Licence commerciale | oui |
| Licence open source | non |
| Poids ouverts | oui |
| Auto-hébergement (DGX Spark) | oui — ≤ 107 Md (Q8/FP8) |
| Modalités | — |
| Fenêtre de contexte | 262 144 |
| KV cache / token (Q8) | 24.6 ko/token |
| Paramètres totaux | 1.73 B |
| Paramètres actifs (session) | 1.73 B |
| Source des paramètres actifs | actifs = total (modèle dense, config HF (LiquidAI/LFM3-1.7B)) ; total : config Hugging Face (LiquidAI/LFM3-1.7B) |
| Curation des paramètres actifs | 2026-11-04 |
| Prix par session | — |
| Tokens / tâche (Data Players) | 1.2 M |
| Tokens / tâche (Terminal-Bench) | 83.3 M (est.) |
| Ratio Terminal-Bench / Data Players | 69.4× |
| Coût / tâche (Terminal-Bench) | — |
| Tokens d'entraînement | 22 000 000 000 000 |
| Coupure de connaissance | — |
| Créé le | 2026-11-02T18:10:22.658834Z |
| Mis à jour le | 2026-11-04T06:04:49.583209Z |
| Source | LLM Stats |
| URL | https://llm-stats.com/models/lfm3-1.7b |
Prix par fournisseur
Statut du modèle (offres + auto-hébergement) : partiel — auto-hébergement DGX Spark (≤ 107 Md).
Toutes sources confondues : catalogue llm-stats, routeurs EUrouter/Cortecs et catalogues directs (OVHcloud, Scaleway, Infomaniak). Les prix sont convertis en USD pour la comparaison ; le prix publié et le taux BCE utilisé sont dans l'infobulle. Les prix inconnus restent indiqués « — ».
Aucune donnée.
Souveraineté qualifiée par offre (modèle × fournisseur) ; survolez le badge pour les critères.
SouverainetéScores par catégorie (top_scores)
| Catégorie | Score |
|---|---|
| chat | 0.90 |
| instruction_following | 0.90 |
| knowledge | 0.50 |
| language | 0.90 |
| long_context | 0.60 |
| math | 0.20 |
| physics | 0.50 |
| reasoning | 0.40 |
| science | 0.50 |
| structured_output | 0.90 |
| frontend_development | 0.40 |
| general | 0.50 |
| agents | 0.50 |
| biology | 0.50 |
| chemistry | 0.50 |
| code | 0.30 |
| communication | 0.90 |
| tool_calling | 0.50 |
| vision | 0.10 |
Scores par benchmark
| Benchmark | Catégorie | Score | Max | Normalisé (API) | Vérifié | Auto-déclaré | Daté du |
|---|---|---|---|---|---|---|---|
| aa-lcr | reasoning | 0.53 | 1.00 | 0.53 | non | oui | 2026-11-04T06:04:49.763229Z |
| aa-omniscience-index | knowledge | 0.50 | 200.00 | — | non | oui | 2026-11-04T06:04:49.750251Z |
| automationbench | agents | 0.41 | 1.00 | 0.41 | non | oui | 2026-11-04T06:04:49.613491Z |
| bfcl-v4 | tool_calling | 0.71 | 1.00 | 0.71 | non | oui | 2026-11-04T06:04:49.623137Z |
| claw-eval | code | 0.69 | 1.00 | 0.69 | non | oui | 2026-11-04T06:04:49.593603Z |
| gpqa | reasoning | 0.63 | 1.00 | 0.63 | non | oui | 2026-11-04T06:04:49.723203Z |
| humanity's-last-exam | reasoning | 0.07 | 1.00 | 0.07 | non | oui | 2026-11-04T06:04:49.733282Z |
| humanity's-last-exam-(with-tools,-text-only) | reasoning | 0.14 | 1.00 | 0.14 | non | oui | 2026-11-04T06:04:49.743316Z |
| ifbench | instruction_following | 0.73 | 1.00 | 0.73 | non | oui | 2026-11-04T06:04:49.700061Z |
| ifeval | chat | 0.97 | 1.00 | 0.97 | non | oui | 2026-11-04T06:04:49.693802Z |
| livecodebench-v6 | reasoning | 0.73 | 1.00 | 0.73 | non | oui | 2026-11-04T06:04:49.670651Z |
| mcp-atlas | reasoning | 0.33 | 1.00 | 0.33 | non | oui | 2026-11-04T06:04:49.632987Z |
| multi-if | chat | 0.89 | 1.00 | 0.89 | non | oui | 2026-11-04T06:04:49.713239Z |
| ruler-128k | long_context | 0.66 | 1.00 | 0.66 | non | oui | 2026-11-04T06:04:49.769947Z |
| scicode | reasoning | 0.28 | 1.00 | 0.28 | non | oui | 2026-11-04T06:04:49.680435Z |
| swe-bench-multilingual | reasoning | 0.23 | 1.00 | 0.23 | non | oui | 2026-11-04T06:04:49.663213Z |
| swe-bench-verified | reasoning | 0.43 | 1.00 | 0.43 | non | oui | 2026-11-04T06:04:49.648480Z |
| tau3-bench | agents | 0.68 | 1.00 | 0.68 | non | oui | 2026-11-04T06:04:49.643105Z |
| terminal-bench-2.1 | reasoning | 0.13 | 1.00 | 0.13 | non | oui | 2026-11-04T06:04:49.667403Z |
| webarena-verified | agents | 0.38 | 1.00 | 0.38 | non | oui | 2026-11-04T06:04:49.599005Z |
Classements
| Catégorie | Rang | Score | Note conservatrice | Benchmarks évalués | Méthode | Daté du | Provenance |
|---|---|---|---|---|---|---|---|
| chat | 1 | 0.90 | 34.07 | 2 | trueskill | 2026-11-04T08:08:43.789169Z | rankings |
| knowledge | 31 | 0.50 | 4.01 | 1 | trueskill | 2026-11-04T08:08:45.139626Z | rankings |
| agents | 97 | — | 18.12 | 6 | trueskill | 2026-11-03 | model_history |
| code | 218 | — | 7.39 | 6 | trueskill | 2026-11-03 | model_history |
Données
Dates de curation (récupération depuis l'API) : models 2026-11-04 · benchmarks 2026-11-04 · scores 2026-11-04 · updates 2026-11-04 · account 2026-11-04 · model_history 2026-11-04 · rankings 2026-11-04