Best Local & Open-Weight LLMs (2026)
Open-weight and self-hostable large language models. Chinese and American labs compared — Qwen, DeepSeek, GLM, Kimi, Llama, Gemma, Mistral, Nemotron, MiniMax, Falcon. Benchmarks, pricing, and hardware requirements (min/mid/max) for running each model locally.
21 tools reviewed
Tier Rankings
Detailed Comparison
| # | Tool | Score | Best For | Price | Free Tier | |
|---|---|---|---|---|---|---|
| 1 |  Qwen (Alibaba) | 8.8 | Developers who want frontier-tier open weights with Apache 2... | Free / $0.12 | Yes | Review |
| 2 |  MiniMax M3 | 8.4 | Agentic coding and tool-use workflows on a budget. Best pric... | Free / $0.30 | Yes | Review |
| 3 |  Gemma 4 (Google) | 8.3 | Developers and businesses who need a permissively licensed m... | Free / $0.14-0.40 | Yes | Review |
| 4 |  IBM Granite 4.0 | 8.2 | Regulated-industry enterprises (healthcare, finance, governm... | Free / Usage-based | Yes | Review |
| 5 |  Kimi K3 (Moonshot) | 8.1 | Agentic coding workflows, tool-use agents, long-horizon repo... | Free / $0 | Yes | Review |
| 6 |  gpt-oss (OpenAI) | 8.1 | Developers who want OpenAI-brand open-weight reasoning model... | Free / $0.15 | Yes | Review |
| 7 |  Arcee Trinity-Large-Thinking | 8.1 | Teams that need a US-made, Apache 2.0, frontier-tier open-we... | Free / $0.90 | Yes | Review |
| 8 |  DeepSeek | 8.0 | Developers and teams who need strong reasoning and coding ca... | Free / $0.15/$0.60 off-peak, $0.30/$1.20 peak | Yes | Review |
| 9 |  GLM / Z.ai (Zhipu AI) | 8.0 | Teams that need frontier-class open weights for agentic codi... | Free / $0.60+ | Yes | Review |
| 10 |  AI21 Jamba2 | 8.0 | Developers building long-context RAG systems (256K context w... | Free / Usage-based | Yes | Review |
| 11 |  Inkling (Thinking Machines Lab) | 8.0 | Teams that want to fine-tune a serious multimodal open-weigh... | Free / 50% off at launch | Yes | Review |
| 12 |  Llama 4 (Meta) | 7.9 | Developers and teams who need a permissively-licensed open-w... | Free / $3-8 | Yes | Review |
| 13 |  Olmo 3 (AI2) | 7.9 | AI researchers doing reproducibility work, training-data stu... | Free / Usage-based | Yes | Review |
| 14 |  LongCat-2.0 (Meituan) | 7.9 | Teams that want frontier-class open weights for audit, fine-... | Free / $0.75 / $2.95 | Yes | Review |
| 15 |  Bonsai 27B (PrismML) | 7.9 | On-device AI builders and privacy-first users who want real ... | Free / $0 | Yes | Review |
| 16 |  Nemotron (Nvidia) | 7.8 | Teams running on Nvidia hardware (TensorRT-LLM, NIM) who nee... | Free / varies | Yes | Review |
| 17 |  StepFun Step 3.7 Flash | 7.8 | Teams building agent systems on Chinese open-weight foundati... | Free / Usage-based | Yes | Review |
| 18 |  Mistral AI | 7.5 | Developers who want cheap, high-quality API access. Also str... | Free / $0.20 | Yes | Review |
| 19 |  Cohere Command A | 7.5 | Mid-size to large enterprises needing a multilingual open-we... | Free / Usage-based | Yes | Review |
| 20 |  Falcon (TII) | 7.1 | Developers who need a genuinely Apache-2.0 small model for o... | Free / varies | Yes | Review |
| 21 | DiffusionGemma (Google) | 6.8 | Developers who need fast local text generation -- autocomple... | Free / undefined | Yes | Review |
All Local & Open-Weight LLMs Reviews
Qwen (Alibaba)
**Qwen-Image-2.1 went open-weight (weights 2026-09-14, blog 2026-09-20)** -- a unified text-to-image and editing model with a 7B visual-generation component, native transparent (RGBA) output, up to 10 reference images and mask-guided local edits -- but under a **research-only licence, not Apache**: commercial use needs a separate licence from Alibaba. Alibaba's open-weights + API family, and August 2026 was its biggest open-release month yet: **Qwen3.8-2.4T-A95B -- the Max-class flagship -- went open-weight on 2026-08-08** (custom Qwen3.8-Max licence), **Qwen3.8-27B** dense landed Apache 2.0 (8/05), and **Qwen3.8-Flash-Next** (8/24 weights, 8/26 blog) previews the Qwen4 architecture: 125B main + 51B n-gram embeddings with only 6B active, served as Qwen3.8-Flash at $0.15/$0.47 per 1M. Qwen 3.7 Max remains the GA API flagship ($2.50/$7.50)
MiniMax M3
**MiniMax-H3's weights are public on Hugging Face (repo 2026-07-28, updated 2026-08-13) under a community licence whose open-weight grant is limited to the US, EU, UK and South Korea**, and **MiniMax Music 3 (2026-08-07)** ships open weights for five-minute full-song generation with a prominent-attribution commercial clause. MiniMax's coding/agent flagship -- M3 (June 1 2026): 1M-token context, MSA sparse attention (>15x decoding speedup at long context), SWE-Bench Pro 59.0%, Terminal-Bench 66.0%. OPEN WEIGHTS LIVE on HuggingFace since June 12 (~428B total / ~23B active, native multimodal, minimax-community license)
Gemma 4 (Google)
Google DeepMind's open-weights model family -- multimodal, 256K context, runs on edge devices **EmbeddingGemma 2 (announced 2026-10-06; weights on Hugging Face since 9/14) adds a 740M-parameter Apache 2.0 embedding model built on Gemma 4 that maps text, code, images, video and audio into one 768-dimensional space, with an 8K context, Matryoshka truncation down to 128 dimensions and about 191 MB of RAM for text-only use on a Pixel 11 Pro.**
IBM Granite 4.0
IBM's enterprise-focused open-weight family -- Granite 4.0 hybrid Mamba-2 + transformer architecture (70-80% memory reduction vs pure transformer), 3B to 32B sizes, Apache 2.0. First open model family to secure ISO 42001 certification. Nano 350M runs on CPU with 8-16GB RAM. 3B Vision variant landed 2026-04-01
Kimi K3 (Moonshot)
Moonshot's 2.8T-parameter Kimi K3 (launched 2026-07-16/17) is the largest open-weight model ever released -- 1M context, multimodal, $3/$15 per 1M via API, ranked best-available on Arena.AI at launch. WEIGHTS SHIPPED ~2026-07-26/27 on Hugging Face (2.8T total / 104B activated, safetensors) under a custom Kimi K3 License, not the Modified MIT of the K2 line
gpt-oss (OpenAI)
OpenAI's FIRST open-weight models -- gpt-oss-120b (single 80GB GPU, near parity with o4-mini on reasoning) and gpt-oss-20b (runs on 16GB edge devices). Apache 2.0. Launched 2025-08-05. gpt-oss-safeguard ships in 2026 as the safety-tuned variant
Arcee Trinity-Large-Thinking
Arcee AI's US-made open-weight frontier reasoning model -- launched 2026-04-01. 398B total params, ~13B active. Sparse MoE (256 experts, 4 active = 1.56% routing). Apache 2.0, trained from scratch. #2 on PinchBench trailing only Claude 3.5 Opus. ~96% cheaper than Opus-4.6 on agentic tasks
DeepSeek
**DeepSeek-V4.1-Flash-1017 (2026-10-17) replaced V4.1-Flash on the web, app and API at unchanged prices** (the deepseek-flash model name now serves it; MIT weights on Hugging Face), with higher scores and shorter outputs, and DeepSeek opened its in-house **Search and Fetch APIs at $0.75 per 1,000 calls per endpoint** -- server-side Search/Fetch calls through DeepSeek Harness, Claude Code and Codex stay free until 2026-10-31 16:00 UTC. Earlier: **DeepSeek-V4.1-Flash shipped 2026-09-10**: a 552B MoE on a new Causal Encoder-Decoder architecture (8B active for input, 16B for output), natively multimodal, MIT-licensed on HF, and priced BELOW the V4-Flash it replaces -- $0.15/$0.60 off-peak, $0.30/$1.20 peak per 1M, with cache-hit input cut to $0.003/$0.006. V4-Flash and V4-Flash-Vision-Exp are retired (aliases route to V4.1-Flash). DeepSeek said V4-Pro would be phased out on 9/14, then **reversed that four days later** -- V4-Pro stays on the API at unchanged $0.66/$1.98 off-peak, $1.32/$3.96 peak until V4.1-Pro lands. V4.1-Flash's vendor benchmarks beat V4-Pro (DeepSWE 74.2, Terminal-Bench 2.1 90.6, GPQA 90.9)
GLM / Z.ai (Zhipu AI)
**GLM 5.2 has a named production deployment at Perplexity: on 2026-09-21 Perplexity disclosed that Perplexity Computer sessions are served by GLM 5.2, post-trained by Perplexity on real user sessions** (21.2% relative cut in live tool-call failures). **GLM-5.3 and GLM-5.3-Flash released 2026-08-25** (weights on Hugging Face): 5.3 is a post-training-only upgrade of the 5.2 base that Z.ai calls 'the most capable open-weights model for coding' (Terminal Bench 2.1 88.2, DeepSWE 66.9, CyberGym 84.5) but moves from MIT to a custom licence; 5.3-Flash (320B/18B active, multimodal) stays MIT at $0.15/$0.50 per 1M.
AI21 Jamba2
AI21 Labs' hybrid SSM-Transformer (Mamba-style) open-weight family -- Jamba2 launched 2026-01-08. Two sizes: 3B dense (runs on phones / laptops) and Jamba2 Mini MoE (12B active / 52B total). Apache 2.0, 256K context, mid-trained on 500B tokens
Inkling (Thinking Machines Lab)
Mira Murati's $12B lab ships its first model (2026-07-15): a 975B/41B-active open-weights MoE that reasons natively over text, images, and audio with a 1M-token context -- positioned not as the strongest model, but as the best starting point for fine-tuning via Tinker
Llama 4 (Meta)
Meta's open-weights family -- Scout (10M context), Maverick (multimodal 400B MoE). NOTE: Meta's frontier work moved to the proprietary Muse Spark line in April 2026; Llama remains downloadable and supported but is effectively in maintenance mode
Olmo 3 (AI2)
Allen Institute for AI's fully-open frontier reasoning models -- Olmo 3 family (2025-11-20) includes 7B and 32B sizes, four variants (Base, Think, Instruct, RLZero). Apache 2.0 with fully open data + checkpoints + training logs. Olmo 3-Think 32B matches Qwen3-32B-Thinking at 6x fewer training tokens
LongCat-2.0 (Meituan)
Meituan's open-source 1.6T-parameter MoE (~48B active) with native 1M-token context, MIT license -- trained entirely on domestic Chinese AI ASICs and revealed as the stealth 'Owl Alpha' model that had been topping OpenRouter
Bonsai 27B (PrismML)
The first 27B-class model that runs on a phone (2026-07-14) -- ternary and 1-bit quantizations of Qwen3.6 27B squeeze a multimodal, tool-calling, 262K-context model into 3.9-5.9GB under Apache 2.0
Nemotron (Nvidia)
Nvidia's open-weights family -- hybrid Mamba-Transformer MoE architecture, optimized for efficient reasoning on Nvidia hardware. Nemotron 3 Ultra (550B total / 55B active) shipped 2026-06-04 as the family flagship, joining Super (120B/12B, March) and Nano
StepFun Step 3.7 Flash
StepFun's (China) agent-focused open-weight family -- Step 3.7 Flash (May 28 2026): 198B sparse MoE vision-language model, ~11B active, 256K context, Apache 2.0, ~400 tok/s, SWE-Bench Pro 56.3. Supersedes Step 3.5 Flash (Feb 2026) as the flagship
Mistral AI
**New 2026-10-15: Timestral, Mistral's first time-series forecasting family (5M, 28M, 123M and 209M parameters, weights on Hugging Face since 10/14 under a non-Apache licence), with the 209M model claimed first on GIFT-Eval, TIME and fev-bench and CPU inference in 10-140 ms.** **Mistral Large 4 ('le Chonk', ML4) entered public preview on 2026-10-06: a 1T-parameter natively multimodal MoE with roughly 50B active parameters, a 1M context window, callable today as mistral-large-4 at $1.36 / $0.14 cached / $4.18 per 1M tokens (batch at half price), open weights promised by the end of October, and vendor-claimed open-weight leads in cyber (Cybench 93%), coding (DeepSWE v1.1 61.7%) and agentic work (AutomationBench 59.9%).** **Mistral opened a Munich hub for Physics AI and Industrial AI on 2026-09-28**, confirming the May 2026 acquisition of Emmi AI (30+ physicists and engineers in computational fluid dynamics, structural mechanics and multi-physics simulation), naming BMW (crash simulation) and Siemens Energy as partners, and committing to **build one gigawatt of European compute capacity by 2030**. **Mistral now powers Firefox Smart Window (2026-09-16)** -- Mozilla's beta AI browsing assistant runs on Mistral models for France and North America, UK and Germany later this year, with zero data retention. **EUR 3B Series D at a EUR 21B+ post-money (9/08, Samsung-led)**; Cloudera partnership (9/10); Medium 3 retired 8/31 in favour of Medium 3.5.
Cohere Command A
Cohere's enterprise-multilingual flagship -- 111B params, 256K context, runs on 2x H100. 23 languages. CC-BY-NC 4.0 on weights (research / non-commercial), commercial requires Cohere enterprise contract. Follow-ups: Command A Reasoning + Command A Vision
Falcon (TII)
UAE's Technology Innovation Institute open-weights family -- Falcon 3 optimized for efficient sub-10B deployment on consumer hardware
DiffusionGemma (Google)
Google DeepMind's experimental open-weights TEXT-DIFFUSION model (June 10, 2026) -- 26B MoE (3.8B active), Apache 2.0, generates 256-token blocks in parallel with bidirectional attention for up to 4x faster output (1,000+ tok/s on H100). Trades some quality vs Gemma 4 for raw speed