All tasks / Text & knowledge / Chat & assistants
Best local AI for chat & assistants
Open-weights models and tools that run entirely on your own hardware, ranked from public benchmarks. Local tools are rated by the best open model they run.
STop ranked: Kimi K3 (Moonshot AI), 1,487 on LMArena Text Arena (overall, style control).99 options listed · data verified 19 Oct 2026
Ranked list
| Tier | Name | Approach | Runs | Rated by | Price | Skill | Compare |
|---|---|---|---|---|---|---|---|
| S |  Kimi K3Moonshot AI | Models | Local | 1,487 on LMArena Text Arena (overall, style control) | ~1600 GB VRAM | — | |
| S |  OllamaOllama | Runtimes & gateways | Local | via Kimi K3 | Free | intermediate | |
| S |  Open WebUIOpen WebUI Inc. | Apps | Local | via Kimi K3 | Free | intermediate | |
| S |  llama.cppggml-org | Runtimes & gateways | Local | via Kimi K3 | Free | advanced | |
| S |  vLLMvLLM project | Runtimes & gateways | Local | via Kimi K3 | Free | advanced | |
| S |  NextChatNextChat | Apps | Local | via Kimi K3 | Free | intermediate | |
| S |  LobeHub (formerly LobeChat)LobeHub | Apps | Local | via Kimi K3 Cloud models: tier S | Free tier · $9.9/mo | intermediate | |
| S |  AnythingLLMMintplex Labs | Apps | Local | via Kimi K3 | Free | beginner | |
| S |  LiteLLMBerriAI | Runtimes & gateways | Local | via Kimi K3 Cloud models: tier S | Free | advanced | |
| S |  Cherry StudioCherryHQ | Apps | Local | via Kimi K3 | Free | beginner | |
| S |  LocalAIEttore Di Giacinto (mudler) | Runtimes & gateways | Local | via Kimi K3 | Free | advanced | |
| S |  exoEXO Labs | Runtimes & gateways | Local | via Kimi K3 | Free | advanced | |
| S |  TextGen (text-generation-webui)oobabooga | Apps | Local | via Kimi K3 | Free | intermediate | |
| S |  LibreChatLibreChat | Apps | Local | via Kimi K3 | Free | advanced | |
| S |  JanMenlo Research | Apps | Local | via Kimi K3 | Free | beginner | |
| S |  ChatboxChatbox AI | Apps | Local | via Kimi K3 | Free tier · $3.99/mo | beginner | |
| S |  KhojKhoj | Apps | Local | via Kimi K3 | Free | intermediate | |
| S |  SGLangSGLang project (LMSYS) | Runtimes & gateways | Local | via Kimi K3 | Free | advanced | |
| S | KTransformers KVCache.AI (Tsinghua MADSys) | Runtimes & gateways | Local | via Kimi K3 | Free | advanced | |
| S |  TensorRT LLMNVIDIA | Runtimes & gateways | Local | via Kimi K3 | Free | advanced | |
| S |  Portkey AI GatewayPortkey | Runtimes & gateways | Local | via Kimi K3 Cloud models: tier S | Free | advanced | |
| S |  XinferenceXorbits | Runtimes & gateways | Local | via Kimi K3 | Free | advanced | |
| S |  LMDeployInternLM (Shanghai AI Laboratory) | Runtimes & gateways | Local | via Kimi K3 | Free | advanced | |
| S |  MLX LMApple (ml-explore) | Runtimes & gateways | Local | via Kimi K3 | Free | advanced | |
| S |  Big-AGIBig-AGI | Apps | Local | via Kimi K3 | Free | intermediate | |
| S |  DeepChatThinkInAI | Apps | Local | via Kimi K3 | Free | beginner | |
| S |  GPUStackGPUStack | Runtimes & gateways | Local | via Kimi K3 | Free | advanced | |
| S |  5irenanbingxyz | Apps | Local | via Kimi K3 | Free | beginner | |
| S |  WitsyKochava Studios | Apps | Local | via Kimi K3 | Free | beginner | |
| S |  LM StudioElement Labs | Apps | Local | via Kimi K3 | Free | beginner | |
| S |  Msty StudioMsty | Apps | Local | via Kimi K3 | Free | beginner | |
| A |  MiMo-V2.6-ProXiaomi | Models | Local | 1,479 on LMArena Text Arena (overall, style control) | ~585 GB VRAM | — | |
| A |  GLM-5.3Z.ai | Models | Local | 1,479 on LMArena Text Arena (overall, style control) | ~425 GB VRAM | — | vs Kimi K3 |
| A |  DeepSeek-V4.1-Flash-1017DeepSeek | Models | Local | 1,475 on LMArena Text Arena (overall, style control) | ~320 GB VRAM | — | |
| A | GLM-5.3-Flash Z.ai | Models | Local | 1,474 on LMArena Text Arena (overall, style control) | ~185 GB VRAM | — | |
| A | DeepSeek-V4-Pro-0813 DeepSeek | Models | Local | 1,465 on LMArena Text Arena (overall, style control) | ~920 GB VRAM | — | |
| A |  Gemma 4 (31B)Google DeepMind | Models | Local | 1,452 on LMArena Text Arena (overall, style control) | ~19 GB VRAM | — | |
| A | MiMo-V2.6-Flash Xiaomi | Models | Local | 1,451 on LMArena Text Arena (overall, style control) | ~178 GB VRAM | — | |
| A |  Step 5 PreviewStepFun | Models | Local | 1,447 on LMArena Text Arena (overall, style control) | ~346 GB VRAM | — | |
| A |  MiniMax-M3MiniMax | Models | Local | 1,441 on LMArena Text Arena (overall, style control) | ~245 GB VRAM | — | |
| A |  InklingThinking Machines Lab | Models | Local | 1,441 on LMArena Text Arena (overall, style control) | ~557 GB VRAM | — | |
| A |  Qwen3.8-27BAlibaba (Qwen) | Models | Local | 1,438 on LMArena Text Arena (overall, style control) | ~17 GB VRAM | — | |
| A |  Mistral Medium 3.5Mistral AI | Models | Local | 1,426 on LMArena Text Arena (overall, style control) | ~75 GB VRAM | — | |
| A |  Nemotron 3 Ultra 550B-A55BNVIDIA | Models | Local | 1,426 on LMArena Text Arena (overall, style control) | ~321 GB VRAM | — | |
| A |  Muse Glimmer 30BMeta | Models | Local | 1,425 on LMArena Text Arena (overall, style control) | ~19 GB VRAM | — | |
| A |  Unsloth (Studio / Desktop app)Unsloth AI | Apps | Local | via DeepSeek-V4.1-Flash-1017 | Free | intermediate | |
| A |  llamafileMozilla.ai | Runtimes & gateways | Local | via GLM-5.3-Flash | Free | intermediate | |
| A |  LLM (CLI)Simon Willison | Runtimes & gateways | Local | via GLM-5.3-Flash Cloud models: tier S | Free | advanced | |
| A |  AIChatsigoden | Runtimes & gateways | Local | via GLM-5.3-Flash | Free | advanced | |
| A |  Page Assistn4ze3m | Apps | Local | via GLM-5.3-Flash | Free | beginner | |
| A |  OsaurusOsaurus | Apps | Local | via GLM-5.3-Flash | Free | intermediate | |
| A |  mistral.rsEric Buehler | Runtimes & gateways | Local | via GLM-5.3-Flash | Free | advanced | |
| A |  Enchantedgluonfield | Apps | Local | via GLM-5.3-Flash | Free | intermediate | |
| A |  llama-swapmostlygeek | Runtimes & gateways | Local | via GLM-5.3-Flash | Free | advanced | |
| A |  LemonadeAMD (lemonade-sdk) | Runtimes & gateways | Local | via GLM-5.3-Flash | Free | intermediate | |
| A |  RamaLamaRamaLama project (containers) | Runtimes & gateways | Local | via GLM-5.3-Flash | Free | advanced | |
| A | Alpaca Jeffser | Apps | Local | via GLM-5.3-Flash | Free | beginner | |
| A |  TabbyAPI (ExLlamaV3)theroyallab | Runtimes & gateways | Local | via GLM-5.3-Flash | Free | advanced | |
| A |  Docker Model RunnerDocker | Runtimes & gateways | Local | via GLM-5.3-Flash | Free | intermediate | |
| A |  BoltAIBoltAI | Apps | Local | via GLM-5.3-Flash Cloud models: tier S | — | beginner | |
| A |  TypingMindTypingMind | Apps | Local | via GLM-5.3-Flash Cloud models: tier S | — | intermediate | |
| A |  Google AI Edge GalleryGoogle | Apps | Local | via Gemma 4 (31B) | Free | beginner | |
| A |  MLC LLMMLC AI | Runtimes & gateways | Local | via Gemma 4 (31B) | Free | advanced | |
| A |  WebLLMMLC AI | Runtimes & gateways | Local | via Gemma 4 (31B) | Free | advanced | |
| A |  PocketPal AIa-ghorbani | Apps | Local | via Gemma 4 (31B) | Free | beginner | |
| A | GenieX Qualcomm | Runtimes & gateways | Local | via Gemma 4 (31B) | Free | advanced | |
| A |  ChatterUIVali-98 | Apps | Local | via Gemma 4 (31B) | Free | intermediate | |
| A |  MaidMobile Artificial Intelligence | Apps | Local | via Gemma 4 (31B) | Free | intermediate | |
| A |  Llamaggml-org and Hugging Face | Apps | Local | via Gemma 4 (31B) | Free | beginner | |
| A |  Locally AIElement Labs (LM Studio) | Apps | Local | via Gemma 4 (31B) | — | beginner | |
| B | Nemotron 3.5 Super VL 120B-A12B NVIDIA | Models | Local | 1,416 on LMArena Text Arena (overall, style control) | ~72 GB VRAM | — | |
| B | Inkling Small Thinking Machines Lab | Models | Local | 1,404 on LMArena Text Arena (overall, style control) | ~159 GB VRAM | — | |
| B |  Trinity Large ThinkingArcee AI | Models | Local | 1,367 on LMArena Text Arena (overall, style control) | ~229 GB VRAM | — | |
| B |  Ling-3.1-flashinclusionAI (Ant Group) | Models | Local | 41 on Artificial Analysis Intelligence Index | ~321 GB VRAM | — | |
| B |  gpt-oss-120bOpenAI | Models | Local | 1,352 on LMArena Text Arena (overall, style control) | ~80 GB VRAM | — | |
| B | Nemotron 3.5 Lightning 30B-A3B NVIDIA | Models | Local | 1,351 on LMArena Text Arena (overall, style control) | ~20 GB VRAM | — | |
| B | Qwen3.8-2.4T-A95B Alibaba (Qwen) | Models | Local | 40 on Artificial Analysis Intelligence Index | ~1370 GB VRAM | — | |
| B | Qwen3.8-Flash-Next Alibaba (Qwen) | Models | Local | 40 on Artificial Analysis Intelligence Index | ~104 GB VRAM | — | |
| C |  Granite 4.2 30BIBM | Models | Local | 1,340 on LMArena Text Arena (overall, style control) | ~18 GB VRAM | — | |
| C | Muse Spark Open (816B-A42B) Meta | Models | Local | 39 on Artificial Analysis Intelligence Index | ~470 GB VRAM | — | |
| C |  Olmo 3.1 32BAllen Institute for AI (Ai2) | Models | Local | 1,329 on LMArena Text Arena (overall, style control) | ~20 GB VRAM | — | |
| C |  Llama 4 MaverickMeta | Models | Local | 1,327 on LMArena Text Arena (overall, style control) | ~230 GB VRAM | — | |
| C | gpt-oss-20b OpenAI | Models | Local | 1,317 on LMArena Text Arena (overall, style control) | ~16 GB VRAM | — | |
| C |  Motif-3Motif Technologies | Models | Local | 34 on Artificial Analysis Intelligence Index | ~181 GB VRAM | — | |
| C |  K2-Horizon-375B-A23BMBZUAI Institute of Foundation Models | Models | Local | 31 on Artificial Analysis Intelligence Index | ~218 GB VRAM | — | |
| C |  Beam 1.0 (501B-A23B)Reflection AI | Models | Local | 30 on Artificial Analysis Intelligence Index | ~287 GB VRAM | — | |
| C | Kimi K2.7 Code Moonshot AI | Models | Local | 26 on Artificial Analysis Intelligence Index | ~587 GB VRAM | — | |
| C | K2-Horizon-MoVA-36B-A4B MBZUAI Institute of Foundation Models | Models | Local | 25 on Artificial Analysis Intelligence Index | ~23 GB VRAM | — | |
| C |  Solar Open2 250BUpstage | Models | Local | 25 on Artificial Analysis Intelligence Index | ~144 GB VRAM | — | |
| C |  A.X K2SK Telecom | Models | Local | 21 on Artificial Analysis Intelligence Index | ~396 GB VRAM | — | |
| C | Nemotron 3.5 Bolt 74B-A9B NVIDIA | Models | Local | 21 on Artificial Analysis Intelligence Index | ~24 GB VRAM | — | |
| C |  K-EXAONE 2.0 (750B-A37B)LG AI Research | Models | Local | 20 on Artificial Analysis Intelligence Index | ~428 GB VRAM | — | |
| C |  LongCat-2.0Meituan (LongCat) | Models | Local | 19 on Artificial Analysis Intelligence Index | ~1014 GB VRAM | — | |
| C |  Command A+ (05-2026)Cohere | Models | Local | 13 on Artificial Analysis Intelligence Index | ~126 GB VRAM | — | |
| C | Ministral 3 14B (2512) Mistral AI | Models | Local | 6 on Artificial Analysis Intelligence Index | ~9 GB VRAM | — | |
| C | Mistral Small 4 (2603) Mistral AI | Models | Local | 11 on Artificial Analysis Intelligence Index | ~69 GB VRAM | — | |
| C | Ring-2.6-1T inclusionAI (Ant Group) | Models | Local | 17 on Artificial Analysis Intelligence Index | ~586 GB VRAM | — | |
| C |  Sarvam-105BSarvam AI | Models | Local | 9 on Artificial Analysis Intelligence Index | ~62 GB VRAM | — | |
| C |  Foundry LocalMicrosoft | Runtimes & gateways | Local | via gpt-oss-20b | Free | intermediate |
Tiers come only from public benchmarks (LMArena Text Arena (overall, style control), Artificial Analysis Intelligence Index); tools without a result of their own inherit the tier of the best model they offer. How we rank.