Can Apple MacBook Air M2 8GB (used) run Llama, Qwen & DeepSeek? 150 models that fit
150 of the 493 models in the Spanvero catalog fit Apple MacBook Air M2 8GB (used)'s 8 GB unified memory (at a sensible quant, 16k context). For each: run it locally ($0 compute + electricity), rent an equivalent GPU ($0 markup, as of 2026-07-09), or pay per-token via your own API key (as of 2026-08-17).
Three honest ways to run each model on Apple MacBook Air M2 8GB (used)
Run it locally: $0 in compute — you pay only electricity (~35 W under load on this Mac). Local is real money, never a fake "$0".
Rent an equivalent GPU: from a $0-markup vendor rate (as of 2026-07-09) — you rent on your own account and pay the vendor directly; we never resell compute.
Skip the box: run the same model through your own API key, paying per million tokens (prices as of 2026-08-17).
What fits Apple MacBook Air M2 8GB (used) (8 GB unified memory)
150 of the 493 notable models in the Spanvero catalog fit Apple MacBook Air M2 8GB (used) at a sensible quant (context capped at 16k for the estimate). Most capable first:
Qwen2.5 Math 7B Instruct (Qwen, 7.6B) — needs ~6 GB at Q4_K_M: run it locally for $0 compute + ~$0.0799/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.16/1M via your own API key (size estimate).
Qwen2.5 Math 7B (Qwen, 7.6B) — needs ~6 GB at Q4_K_M: run it locally for $0 compute + ~$0.0799/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.16/1M via your own API key (size estimate).
OLMoE 1B 7B 0125 Instruct (allenai, 6.9B) — needs ~6 GB at Q4_K_M: run it locally for $0 compute + ~$0.0739/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.16/1M via your own API key (size estimate).
OLMoE 1B 7B 0924 (allenai, 6.9B) — needs ~6 GB at Q4_K_M: run it locally for $0 compute + ~$0.0739/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.16/1M via your own API key (size estimate).
AFM 4.5B (arcee-ai, 4.6B) — needs ~5 GB at Q4_K_M: run it locally for $0 compute + ~$0.0534/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.14/1M via your own API key (size estimate).
Agents A1 4B (InternScience, 4.5B) — needs ~6 GB at Q4_K_M: run it locally for $0 compute + ~$0.0525/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.14/1M via your own API key (size estimate).
Qwen3Guard Gen 4B (Qwen, 4.4B) — needs ~6 GB at Q4_K_M: run it locally for $0 compute + ~$0.0516/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.14/1M via your own API key (size estimate).
Qwen3 4B (Qwen, 4B) — needs ~5 GB at Q4_K_M: run it locally for $0 compute + ~$0.0478/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.13/1M via your own API key (size estimate).
Qwen3 4B Instruct 2507 (Qwen, 4B) — needs ~5 GB at Q4_K_M: run it locally for $0 compute + ~$0.0478/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.13/1M via your own API key (size estimate).
Rio 3.0 Open Mini (prefeitura-rio, 4B) — needs ~5 GB at Q4_K_M: run it locally for $0 compute + ~$0.0478/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.13/1M via your own API key (size estimate).
NVIDIA Nemotron 3 Nano 4B BF16 (nvidia, 4B) — needs ~6 GB at Q4_K_M: run it locally for $0 compute + ~$0.0478/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.13/1M via your own API key (size estimate).
Qwen3 4B Base (Qwen, 4B) — needs ~5 GB at Q4_K_M: run it locally for $0 compute + ~$0.0478/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.13/1M via your own API key (size estimate).
Qwen3 4B Thinking 2507 (Qwen, 4B) — needs ~5 GB at Q4_K_M: run it locally for $0 compute + ~$0.0478/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.13/1M via your own API key (size estimate).
typhoon2.5 qwen3 4b (typhoon-ai, 4B) — needs ~5 GB at Q4_K_M: run it locally for $0 compute + ~$0.0478/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.13/1M via your own API key (size estimate).
Phi-3.5-mini Instruct (Microsoft, 3.8B) — needs ~5 GB at Q4_K_M: run it locally for $0 compute + ~$0.0459/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.13/1M via your own API key (size estimate).
Phi 4 mini instruct (microsoft, 3.8B) — needs ~6 GB at Q4_K_M: run it locally for $0 compute + ~$0.0459/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.22/1M via your own API key (last-known).
Phi tiny MoE instruct (microsoft, 3.8B) — needs ~4 GB at Q4_K_M: run it locally for $0 compute + ~$0.0459/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.13/1M via your own API key (size estimate).
Phi 3 mini 4k instruct (microsoft, 3.8B) — needs ~5 GB at Q4_K_M: run it locally for $0 compute + ~$0.0459/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.13/1M via your own API key (size estimate).
Nemotron Labs Diffusion 3B Base (nvidia, 3.8B) — needs ~4 GB at Q4_K_M: run it locally for $0 compute + ~$0.0459/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.13/1M via your own API key (size estimate).
Phi 4 mini reasoning (microsoft, 3.8B) — needs ~6 GB at Q4_K_M: run it locally for $0 compute + ~$0.0459/1M in electricity, rent NVIDIA RTX 3060 12GB from $0.06/hr ($0 markup), or ~$0.13/1M via your own API key (size estimate).
The honest cost of owning Apple MacBook Air M2 8GB (used)
Street price $499.00 (as of 2026-07-09; Swappa live listings + Mac of All Trades direct ($507.99 Good) + RefurbMe eBay reading ($500 Good 256GB), retrieved 2026-07-09 (fair-grade floor $419)) — amortized over 3 years that's ~$0.4557/day whether or not you're generating.
Electricity: ~35 W under sustained inference at $0.1883/kWh (EIA Electric Power Monthly Table 5.6.A — U.S. residential average, Apr 2026, as of 2026-07-02) — the per-1M-token figures above already include this at each model's speed.
Straight talk: for the small models a 8 GB unified memory box runs, hosted APIs are often cheaper per token. Own local for privacy, offline use, and unlimited runs — not to save money on tokens.
Too big for Apple MacBook Air M2 8GB (used) — rent or use an API instead
These need more than the 8 GB unified memory on this Mac. Closest first — you can still run them on a rented GPU ($0 markup) or via your own API key:
Nemotron Labs Diffusion 8B Base (nvidia, 8.5B) — needs ~7 GB; rent NVIDIA RTX 3060 12GB from $0.06/hr, or ~$0.17/1M via your own API key (size estimate).
LFM2.5 8B A1B (LiquidAI, 8.5B) — needs ~7 GB; rent NVIDIA RTX 3060 12GB from $0.06/hr, or ~$0.17/1M via your own API key (size estimate).
LFM2 8B A1B (LiquidAI, 8.3B) — needs ~7 GB; rent NVIDIA RTX 3060 12GB from $0.06/hr, or ~$0.17/1M via your own API key (size estimate).
MiniCPM4.1 8B (openbmb, 8.2B) — needs ~7 GB; rent NVIDIA RTX 3060 12GB from $0.06/hr, or ~$0.17/1M via your own API key (size estimate).
MiniCPM4 8B (openbmb, 8.2B) — needs ~7 GB; rent NVIDIA RTX 3060 12GB from $0.06/hr, or ~$0.17/1M via your own API key (size estimate).
granite 3.0 8b instruct (ibm-granite, 8.2B) — needs ~7 GB; rent NVIDIA RTX 3060 12GB from $0.06/hr, or ~$0.17/1M via your own API key (size estimate).
A short email of real AI price moves, straight from the daily log — no hype. We're collecting the list now; the first issue goes out when it opens. Unsubscribe with one click.