Ranking updated on 09/10/2026
The MacBook Air M2 (8 / 16 / 24 GB, 100 GB/s) remains highly usable in 2026 for local AI thanks to modern Q4 quantizations. Target 3-8B without a fan.
MacBook Air M2: purchasing alternative available for local AI — MacBook Pro M5 Pro — 24 GB / 1 TB:
Which PC should you choose for your budget? Our picks from €800 to €3,500 →
Affiliate links — BestLLMfor may earn a commission on purchases, at no extra cost to you, which does not influence the ranking (established independently). As an Amazon Associate, BestLLMfor earns from qualifying purchases.
4B effective multimodal (text+image+audio). 140 languages. For laptops and edge devices.
ollama run gemma4:e4b
Dense 8B Apache 2.0, 12 languages including FR, 131k context, GQA 32Q/8KV. MMLU 73.84, HumanEval 85.37. Released April 29, 2026.
# HuggingFace : ibm-granite/granite-4.1-8b
Granite 4.2 8B (IBM): dense Apache 2.0, 128k context, ~4.6 GB Q4 VRAM. Multilingual chat, coding, and reasoning for the enterprise.
ollama pull granite4.2
SFT “thinking” fine-tune of OLMo 3 7B: step-by-step reasoning, 16k context, ~4.2 GB VRAM in Q4. 100% open, Apache 2.0 license.
# HuggingFace : zimplex/olmo3-7b-think-sft-eosfix-16k-3ep-euc
GLM 5.3 (Zhipu): dense 7B specialized in code and reasoning, 128k context, ~4.1 GB VRAM in Q4. Lightweight, runs on a 6–8 GB GPU, MIT license.
ollama pull glm-5.3
Dense 7B 100% open (weights + data + code). Complete transparency for research.
ollama run olmo-3:7b
Hybrid thinking/fast mode. 119 languages, 32k native (131k via YaRN).
ollama run qwen3:8b
| Rank | Model | Params | Q4 VRAM | Context | License | On Apple M2 (16 GB) |
|---|---|---|---|---|---|---|
| #1 | Gemma 4 E4B | 4B | 10 GB | 128 000 | Apache 2.0 | 14 tok/s · Q4_K_M |
| #2 | Granite 4.1 8B Instruct | 8B | 5 GB | 131 072 | Apache 2.0 | 12 tok/s · Q8 |
| #3 | Granite 4.2 8B | 8B | 4.6 GB | 128 000 | Apache 2.0 | 32 tok/s · Q8 |
| #4 | OLMo 3 7B Think (SFT) | 7B | 4.2 GB | 16 000 | Apache 2.0 | 32 tok/s · Q8 |
| #5 | GLM 5.3 7B | 7B | 4.1 GB | 128 000 | MIT | 32 tok/s · Q8 |
| #6 | OLMo 3 7B | 7B | 5 GB | 8 192 | Apache 2.0 | 12 tok/s · Q8 |
| #7 | Qwen 3 8B | 8B | 5 GB | 131 072 | Apache 2.0 | 12 tok/s · Q8 |
Local AI on your Mac, fully explored: unified memory, MLX vs. GGUF, the right model for your chip, Ollama and LM Studio tuned for Apple Silicon.
Free memo
Get the memo VRAM → best coding model → Ollama command (one screen, copy and paste). Then switch to the Copilote Local kit for a setup that actually works.
The Local Copilot kit — the Ollama + Cline + Aider configs are ready to paste, with tuned Modelfiles, troubleshooting, and lifetime online access →No spam. Unsubscribe in 1 click. Your data stays with us (never resold).
Your card → the best coding model to run locally, and the exact Ollama command:
| Your VRAM | Typical GPUs / Macs | Recommended coding model | Command Ollama |
|---|---|---|---|
| 8 GB | RTX 4060 / 3060 · M1-M2 16 GB | Qwen 3.5 9B (Q4, 6.6 GB — 256k context) | ollama run qwen3.5:9b |
| 12 GB | RTX 3060 12 GB / 4070 / 5070 | Qwen 3.5 9B (Q8, 11 GB) or Gemma 4 12B (7.6 GB) | ollama run qwen3.5:9b-q8_0 |
| 16 GB | RTX 5070 Ti / 4080 / 5080 · RX 9070 XT · M4 24 GB | Devstral 24B (Q4, 14 GB) — coding-agent specialist | ollama run devstral:24b |
| 24 GB | RTX 3090 / 4090 · RX 7900 XTX · M4 Pro 48 GB | Qwen 3.8 27B (Q4, 18 GB) — the “close to Copilot” option | ollama run qwen3.8:27b |
| 32 GB | RTX 5090 | Qwen 3.6 35B-A3B (Q4, 23 GB) — fast MoE | ollama run qwen3.6:35b |
| 48 GB+ | Mac M4 Max 64 GB · M2 Ultra 128 GB | Qwen3-Coder 30B-A3B (Q8, 32 GB — 256k context) | ollama run qwen3-coder:30b-a3b-q8_0 |
-base: ollama run qwen2.5-coder:7b-base — it’s still the reference for this specific use case. ⚠️ Qwen 3.8: its reasoning is set very high by default and it “overthinks” simple requests — lower it to low (or turn it off) on first launch. ⚠️ License trap: Codestral 22B = Mistral Non-Production License → prohibited for coding at work. Qwen 3.5/3.8, Gemma 4, and Devstral are Apache 2.0. 💡 Running out of memory? Keep ~1.5 GB of VRAM free for context, or drop down one quantization level.🔌 To connect it to VS Code: Cline (multi-file agent), Aider (CLI) or Tabby/Twinny (FIM autocomplete) — they all connect to Ollama locally. The kit Local Copilot — ready-to-paste configs + tested setup — is available: /copilote-local.
Filter: 1-13B with Q4_K_M fitting under 11 GB. Bonus for 3-8B (peak Air M2) and an additional bonus for ≤ 3B (fast even without the M4 Neural Engine).
Criteria considered:
The scoring is fully transparent: see our methodology for details on VRAM/tokens/sec calculations.
Is a 16 GB MacBook Air M2 still viable in 2026?
Yes — Mistral 7B Q4, Qwen 3 8B Q4, and Phi-4 Mini run at 18-25 tokens/sec. Sufficient for personal chat / RAG / light code completion. See the MacBook Air M2 guide.
M2 vs M3 vs M4 on the same 7B model?
M2 ≈ 22 tok/s, M3 ≈ 26 tok/s, M4 ≈ 30 tok/s on Mistral 7B Q4_K_M. The difference is noticeable, but none is slow. On battery, M2 drains slightly faster (GPU efficiency).
Air M2 24 GB: which large models?
Mistral Nemo 12B Q4 (~7 GB) or Qwen 3 14B Q4 (~8 GB) run at 12–18 tokens/sec. Avoid 30B+ models, even MoE: memory bandwidth is too limited.
Air M2 8 GB: seriously feasible?
Only with Phi-4 Mini 3.8B Q4 or Llama 3.2 3B Q4. Maximum context: 2k–4k. This is a stopgap: 16 GB remains the practical minimum for serious LLMs.
Prices in euros (€) are French market prices including VAT, as checked by BestLLMfor. US prices differ: the Amazon buttons show the current US price.