LFM2.5-VL-3B
LFM2.5-VL-3B is Liquid AI's open-weight vision-language model (~3.12B parameters; HF safetensors total 3,123,483,888) for on-device and edge image-text→text. Non-reasoning: answers directly for low latency. Builds on the LFM2.5-2.6B language backbone with a SigLIP2 400M NaFlex vision encoder; improves screen/UI understanding, grounding, function calling, and multi-image input over LFM2-VL-3B. Pre-trained on ~34T tokens; 128K vocabulary; 32,768-token context. Day-one support for llama.cpp, MLX, vLLM, SGLang, and ONNX. License: LFM Open License v1.0 (lfm1.0). Catalog benches include MMStar, MME (Liquid 0–100 normalize of 0–2800), RealWorldQA, SimpleVQA, MMBench-V1.1, MM-IF-Eval, MathVista-Mini, MMMU-Pro, MMMU (val), ChartQA, DocVQA, OCRBench, OCRBench-V2 (en), TextVQA, RefCOCO-avg, BLINK, MuirBench, HallusionBench, POPE, IFEval, IFBench, Multi-IF, and BFCL-v4. No catalog id for ScreenSpot-v2 (80.7 avg), CountBenchQA, SEED-Bench (image), MMMB, Multilingual MMBench, LogicVista, InfographicVQA, or ToolSandbox — recorded here only. Sources: https://www.liquid.ai/blog/lfm2-5-vl-3b · https://huggingface.co/LiquidAI/LFM2.5-VL-3B. No hosted provider pricing row (self-host open weights; OpenRouter has only liquid/lfm-2.5-2.6b:free for the text sibling).
Identité
| Organisation | Liquid AI |
|---|---|
| Pays | — |
| Famille | — |
| Type | llm |
| Date de sortie | 2026-08-12 |
| 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 | 32 768 |
| KV cache / token (Q8) | 30.7 ko/token |
| Paramètres totaux | 3.12 B |
| Paramètres actifs (session) | 3.12 B |
| Source des paramètres actifs | actifs = total (modèle dense, config HF (LiquidAI/LFM2.5-VL-3B)) ; total : config Hugging Face (LiquidAI/LFM2.5-VL-3B) |
| 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 | 34 000 000 000 000 |
| Coupure de connaissance | — |
| Créé le | 2026-08-12T18:43:51.674679Z |
| Mis à jour le | 2026-11-04T06:04:49.420862Z |
| Source | LLM Stats |
| URL | https://llm-stats.com/models/lfm-2.5-vl-3b |
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 | 0.70 |
| image_to_text | 0.70 |
| instruction_following | 0.60 |
| language | 0.60 |
| math | 0.70 |
| multimodal | 0.70 |
| reasoning | 0.60 |
| safety | 0.90 |
| spatial_reasoning | 0.70 |
| structured_output | 0.70 |
| general | 0.60 |
| grounding | 0.90 |
| healthcare | 0.50 |
| 3d | 0.60 |
| agents | 0.30 |
| communication | 0.60 |
| tool_calling | 0.30 |
| vision | 0.70 |
Scores par benchmark
| Benchmark | Catégorie | Score | Max | Normalisé (API) | Vérifié | Auto-déclaré | Daté du |
|---|---|---|---|---|---|---|---|
| bfcl-v4 | tool_calling | 0.33 | 1.00 | 0.33 | non | oui | 2026-11-04T06:04:49.570299Z |
| blink | vision | 0.61 | 1.00 | 0.61 | non | oui | 2026-11-04T06:04:49.523262Z |
| chartqa | reasoning | 0.81 | 1.00 | 0.81 | non | oui | 2026-11-04T06:04:49.476089Z |
| docvqa | vision | 0.91 | 1.00 | 0.91 | non | oui | 2026-11-04T06:04:49.479638Z |
| hallusion-bench | vision | 0.47 | 1.00 | 0.47 | non | oui | 2026-11-04T06:04:49.543182Z |
| ifbench | instruction_following | 0.26 | 1.00 | 0.26 | non | oui | 2026-11-04T06:04:49.559660Z |
| ifeval | chat | 0.82 | 1.00 | 0.82 | non | oui | 2026-11-04T06:04:49.552355Z |
| mathvista-mini | math | 0.69 | 1.00 | 0.69 | non | oui | 2026-11-04T06:04:49.463068Z |
| mm-if-eval | multimodal | 0.61 | 1.00 | 0.61 | non | oui | 2026-11-04T06:04:49.457305Z |
| mmbench-v1.1 | vision | 0.81 | 1.00 | 0.81 | non | oui | 2026-11-04T06:04:49.453195Z |
| mme | vision | 0.73 | 1.00 | 0.73 | non | oui | 2026-11-04T06:04:49.433126Z |
| mmmu-(val) | vision | 0.48 | 1.00 | 0.48 | non | oui | 2026-11-04T06:04:49.473055Z |
| mmmu-pro | vision | 0.30 | 1.00 | 0.30 | non | oui | 2026-11-04T06:04:49.469693Z |
| mmstar | vision | 0.63 | 1.00 | 0.63 | non | oui | 2026-11-04T06:04:49.424958Z |
| muirbench | vision | 0.58 | 1.00 | 0.58 | non | oui | 2026-11-04T06:04:49.533405Z |
| multi-if | chat | 0.59 | 1.00 | 0.59 | non | oui | 2026-11-04T06:04:49.563122Z |
| ocrbench | vision | 0.84 | 1.00 | 0.84 | non | oui | 2026-11-04T06:04:49.483009Z |
| ocrbench-v2-(en) | vision | 0.47 | 1.00 | 0.47 | non | oui | 2026-11-04T06:04:49.493137Z |
| pope | vision | 0.89 | 1.00 | 0.89 | non | oui | 2026-11-04T06:04:49.548117Z |
| realworldqa | vision | 0.73 | 1.00 | 0.73 | non | oui | 2026-11-04T06:04:49.442976Z |
| refcoco-avg | vision | 0.88 | 100.00 | 0.88 | non | oui | 2026-11-04T06:04:49.513196Z |
| simplevqa | vision | 0.35 | 100.00 | 0.35 | non | oui | 2026-11-04T06:04:49.447282Z |
| textvqa | vision | 0.84 | 1.00 | 0.84 | non | oui | 2026-11-04T06:04:49.503262Z |
Classements
| Catégorie | Rang | Score | Note conservatrice | Benchmarks évalués | Méthode | Daté du | Provenance |
|---|---|---|---|---|---|---|---|
| agents | 198 | — | -5.31 | 1 | trueskill | 2026-10-19 | 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