Local AI by Editorial Team

Best Local LLM for Everyday Tasks (Chat, Writing, Summarizing)

Which open-weight LLM should you actually run for everyday chat, writing, and summarizing? A tier-by-tier guide from 8 GB laptops to 32 GB desktops.

“Best local LLM” used to mean whichever 7B you could squeeze into 8 GB. The 2026 field is wider: a 1B model on a thin laptop, a 4B on an 8 GB card, a 24B on a 24 GB card, and a 70B on a workstation. Download sizes, context windows, and licences below come from the Ollama library pages and Hugging Face model cards (checked 2026-09-27 and rechecked 2026-09-29). Sizes are the default 4-bit (Q4_K_M) downloads, in GB as Ollama lists them; the memory needed at run time is higher once the context (KV cache) is added.

TL;DR

  • 8 GB laptops and Steam Decks: llama3.2:3b (2.0 GB) for the lightest path, qwen3:4b (2.5 GB) if you can spare a little more RAM and want an Apache-2.0 licence.
  • 16 GB tier: qwen3:8b (5.2 GB, 40K context) or mistral-small:24b (14 GB, 32K context) if you have headroom; with a long context it is a tight fit on a 16 GB card.
  • 24 GB tier: mistral-small:24b is the practical sweet spot. qwen3:14b (9.3 GB, 40K context) is the lower-memory alternative with 100+ language support.
  • 32 GB+ tier: qwen3:32b (20 GB) is the largest dense Apache-2.0 Qwen3 and fits this tier. llama3.3:70b (43 GB) is the next step up if you accept Meta’s licence terms.
  • License picks matter as much as size. Apache-2.0 (Qwen3, Mistral Small 24B) and MIT (Phi-4) are the safe picks for commercial or redistributable work. Llama 3.2/3.3 and Gemma 3 have custom licences that require acceptance and have restrictions on use.

What “everyday tasks” actually covers

Everyday is the opposite of agentic coding and large-context research. It is a chat that does not lose the thread of the conversation, a paragraph rewrite that does not sound like a Wikipedia stub, and a long document or article that can be summarized without the model losing the first page by the time it reaches the last. Three properties drive almost every recommendation below:

  1. Instruction following - whether the model reads your prompt and does what you asked rather than guessing.
  2. Prose quality - whether the output reads naturally instead of stilted or repetitious.
  3. Long-context retention - whether the model can hold a 10-page article or a 30-message thread without losing the start.

Those three collapse into a simple hardware ladder, because every size jump buys more of all three at once.

The 2026 open-weight field for chat and writing

Five model families cover the realistic choices for an everyday local assistant. Two are Apache-2.0 (Qwen3, Mistral Small), one is MIT (Phi-4), one is Meta’s Llama 3.2/3.3 Community License, and one is Google’s Gemma license. The licence decision is made before the size decision: if your work is commercial or redistributable, narrow to Apache-2.0 and MIT first.

Qwen3 (Apache-2.0, Alibaba)

The Qwen3 family ships at 0.6B, 1.7B, 4B, 8B, 14B, 30B (MoE), 32B, and 235B (MoE), with a 40K-token context window on most tags and 256K on the 4B, 30B, and 235B tags. The original Hugging Face card for Qwen3-4B lists 32,768 tokens natively, extendable to 131,072 with YaRN; Ollama’s 4B tag was updated later and lists 256K, so check which build you pulled. The Ollama library page advertises 100+ languages and dialects, switching between a “thinking” mode for reasoning and a “non-thinking” mode for chat, and an Apache-2.0 licence across the family.

For everyday tasks the relevant Qwen3 sizes are qwen3:1.7b (1.4 GB, 2.03B parameters, Q4_K_M), qwen3:4b (2.5 GB, 4.02B parameters), qwen3:8b (5.2 GB), and qwen3:14b (9.3 GB). The 30B (19 GB download) and 235B (142 GB) are MoE checkpoints; the 30B belongs in the 24 GB tier and the 235B needs a multi-GPU workstation, which is out of scope for everyday hardware.

Gemma 3 (Gemma Terms of Use, Google)

The Gemma 3 family ships at 270M, 1B, 4B, 12B, and 27B, with 128K context for the 4B, 12B, and 27B sizes and 32K for the 1B. The 270M and 1B are text-only; the rest are multimodal (text and image input). Gemma 3 is gated: the Ollama tag page carries the “Gemma Terms of Use” (last modified February 21, 2024), and the Hugging Face card requires you to agree to Google’s usage licence before access.

The relevant sizes for everyday local use are gemma3:1b (806 MB, 32K context, text-only), gemma3:4b (3.4 GB, 128K context, multimodal), and gemma3:12b (8.1 GB, 128K context, multimodal). The 4B and 12B are the sizes that actually compete with Qwen3 at the same VRAM tier; the 27B (17 GB) is the 24 GB+ tier.

Mistral Small (Apache-2.0, Mistral AI)

mistral-small:24b is the headline tag: 14 GB, 23.6B parameters, Q4_K_M, 32K context, Apache-2.0. The Hugging Face Mistral-Small-24B-Instruct-2501 card confirms the Apache-2.0 licence and 32K context. The card lists a Tekken tokenizer with a 131k vocabulary and support for “dozens of languages”, naming English, French, German, Spanish, Italian, Chinese, Japanese, Korean, Portuguese, Dutch, and Polish.

Mistral’s plain mistral:7b is still a viable 4.4 GB tag at 32K context for the 16 GB tier; the older Mistral-7B-Instruct-v0.3 card lists Apache-2.0. The 7B is a workhorse more than a frontier model: it is the one to pick when a 24B will not fit and you want a model that is not Qwen3 or Llama.

Llama 3.2 and 3.3 (Llama Community License, Meta)

Meta’s open-weight chat line is split across two licences: the Llama 3.2 Community License covers the 1B and 3B tags, and the Llama 3.3 Community License covers the 70B. Both are custom commercial licences that require acceptance; companies above 700 million MAUs must request a separate licence from Meta. Both 3.2 3B and 3.3 70B list 128K context.

The realistic Ollama tags for everyday local use are llama3.2:3b (2.0 GB, 128K context, 8 languages) for the low end and llama3.3:70b (43 GB) for the high end. The 70B is not the best fit for everyday tasks: it is a frontier model on a workstation, not a casual chat model on a 24 GB card.

Phi-4 (MIT, Microsoft Research)

microsoft/phi-4 is a 14B dense decoder-only Transformer, MIT-licensed, with 16K context, released December 12, 2024. Microsoft’s card reports strong reasoning numbers (vendor-reported: MMLU 84.8, GPQA 56.1, MATH 80.4, HumanEval 82.6), a low factual-recall score (SimpleQA 3.0), and a mostly English training mix (multilingual data is about 8%). For English chat and writing it sits in the same size class as qwen3:14b. For multilingual or summarization-heavy workloads, Qwen3 or Mistral Small is the safer pick.

Pick by hardware tier

Under 4 GB of usable VRAM (8 GB laptops, Steam Deck, integrated graphics)

The realistic options are llama3.2:1b (1.3 GB), qwen3:1.7b (1.4 GB, 2.03B params, Apache-2.0), gemma3:1b (806 MB, 32K context, text-only), and llama3.2:3b (2.0 GB). Pick by licence first:

  • Apache-2.0 only: qwen3:1.7b. It has a 40K context window and the Qwen3 family’s 100+ language support.
  • Meta licence acceptable: llama3.2:3b (2.0 GB, 128K context, 8 officially supported languages) is the larger model at this tier.
  • Gemma licence acceptable: gemma3:1b (806 MB) is the smallest download of the three, with a 32K context window and text-only input.

8 GB usable VRAM (entry-level discrete GPUs, M1/M2 base)

qwen3:4b (2.5 GB, Apache-2.0; context as above) is a sensible default at this tier. gemma3:4b (3.4 GB, 128K context, multimodal) is the alternative if you want image input; it is 0.8 GB larger.

16 GB usable VRAM (RTX 4060 Ti, M2 Pro, M3 Pro)

qwen3:8b (5.2 GB, 40K context, Apache-2.0) and mistral:7b (4.4 GB, 32K context, Apache-2.0) are the realistic picks. Both fit a 16 GB card with room for a long context. Both are Apache-2.0. qwen3:8b has the longer context window and the broader language list; mistral:7b is the pick if you specifically want a non-Qwen3, non-Llama model.

24 GB usable VRAM (RTX 4090, M3 Max, M4 Pro)

This is the tier where everyday chat stops feeling like a demo. Three real options:

  • mistral-small:24b (14 GB, 23.6B params, Q4_K_M, 32K context, Apache-2.0) - the recommended default: the largest model of the three, with dozens of languages and a 32K window for everyday documents and long threads.
  • qwen3:14b (9.3 GB, 40K context, Apache-2.0) - about two thirds of the download size, 100+ languages, and a slightly longer context window.
  • gemma3:12b (8.1 GB, 128K context, multimodal) - the Gemma option, gated licence, multimodal input.

mistral-small:24b is the largest of the three; qwen3:14b is the pick if memory headroom or language coverage matters more than model size.

32 GB+ usable VRAM (workstations, dual GPUs)

qwen3:32b (20 GB, 40K context, Apache-2.0) is the largest dense Qwen3. qwen3:30b (19 GB, 256K context, MoE) is the alternative that opens up long-context work. llama3.3:70b (43 GB, 128K context, 8 languages) is the frontier chat model at this tier but requires Meta’s licence and most of a 48 GB card to run at full Q4.

For everyday tasks, qwen3:32b is the practical ceiling. Going higher buys benchmarks more than it buys a better chat experience.

What the licences actually let you do

The licence filter matters more than the size filter for anyone shipping product, redistributing weights, or using AI inside a company above Meta’s 700M-MAU threshold. The Apache-2.0 models (Qwen3, Mistral Small 24B, Mistral 7B) and the MIT model (Phi-4) impose no usage restrictions; their conditions are about keeping licence and attribution notices. Llama 3.2/3.3 require acceptance of Meta’s Llama Community License and carve out the 700M-MAU restriction. Gemma 3 requires acceptance of Google’s Gemma Terms of Use and is gated on Hugging Face.

For everyday personal use, any of the five works. For commercial or redistributable work, narrow to Apache-2.0 and MIT first; that is Qwen3, Mistral Small, Mistral 7B, and Phi-4. The open-model licences hub has the wider matrix.

A few things that are not on the cards

Tokens per second are not on the cards. None of the cited pages publishes speed figures for consumer hardware, and this guide does not either. Run the model on the hardware in front of you and trust your own number.

The advertised context window is not what fits in VRAM. The context-window article covers the KV-cache math; the takeaway is that a 24 GB card with mistral-small:24b (32K context) holds the full advertised window, while a 16 GB card with qwen3:4b cannot hold 131K tokens. The arithmetic, with an fp16 KV cache: Mistral Small 24B has 40 layers and 8 KV heads of dimension 128, so each token costs 2 x 40 x 8 x 128 x 2 bytes = 163,840 bytes, and 32,768 tokens take about 5.4 GB (5.0 GiB) on top of the 14 GB of weights. Qwen3-4B (36 layers, 8 KV heads of 128) costs 147,456 bytes per token, so 131,072 tokens need about 19.3 GB (18.0 GiB) of cache alone, more than a 16 GB card holds.

Thinking mode is not free. Several Qwen3 sizes can run in a “thinking” mode that improves reasoning but spends tokens. The thinking-mode cost article has the breakdown; for everyday chat the default is non-thinking.

The Bottom Line

For everyday chat, writing, and summarizing, the size ladder from 8 GB laptops to 32 GB workstations is short: qwen3:1.7b or llama3.2:3b at the bottom, qwen3:4b in the middle, qwen3:8b or mistral:7b at 16 GB, mistral-small:24b at 24 GB, and qwen3:32b at 32 GB. The two non-obvious calls are the licence filter (narrow to Apache-2.0 for commercial work) and the multimodal question (Gemma 3 at 4B and above is the only image-input choice on this list, under Google’s custom terms). Pick the model whose licence you can live with, match the size to your hardware, and verify the context window on the hardware in front of you rather than trusting the advertised number.