Mac Mini M4 16 GB

The $499 entry Mac is gone — Apple discontinued the base config May 1.

Apple discontinued the $599 Mac mini base config on May 1, 2026 and raised the floor to $799 with 512 GB. The 16 GB / 256 GB SKU only survives on Amazon residuals and eBay. If you can find one near $499, the 8B-class story still holds; otherwise the math has shifted toward the 24 GB M4 Pro.

The decision in five lines

The call: Consider — The $499 entry Mac is gone — Apple discontinued the base config May 1.
Best for: Entry Mac
Runs well: Qwen 3.5 4B · Qwen 3.5 4B + tight RAG · SANA-0.6B (non-commercial)
Watch out: Apple discontinued the $599 base config (May 1, 2026); the 16 GB / 256 GB SKU is no longer listed on Apple's site. New floor is $799 (512 GB). Tim Cook flagged "several months" of supply constraint on the May 1 earnings call — DRAM shortage drove the cull.
Evidence: Measured · last verified June 2026

16: GB UNIFIED
120: GB/S BANDWIDTH
65: W PEAK (4W IDLE)
$799: NEW FLOOR (512GB)

What fits at this tier

Fits Llama 3.1 8B Q4 comfortably (~5 GB, plenty of headroom). 14B Q4 is tight — requires `OLLAMA_MAX_LOADED_MODELS=1` and closing browser tabs. 20B+ MoE (gpt-oss-20b, 30B-A3B) does not fit at any usable quant. Throughput: 20–40 tok/s on 8B Q4.

CODING

Qwen 3.5 4B 4B dense with 262K context; surprisingly coherent for its size.

CHAT / GENERAL

Qwen 3.5 4B 4B dense with 262K context and native multimodal.

DOCS & RETRIEVAL

Qwen 3.5 4B + tight RAG 4B plus tight chunking; keep context windows small.

IMAGE

SANA-0.6B (non-commercial) 0.6B params; <1s per 1024² on a 16GB laptop GPU; weights are NVIDIA NSCL v2 (non-commercial).

AGENTS

Ministral 3 3B Smallest Ministral with reasoning + tool use.

VOICE

Kokoro-82M (Apache 2.0) Community daily driver for English TTS; CPU-real-time at 82M params; v1.0 with 8 languages and 54 voices. No voice cloning.

The call

Buy it at $799 (512 GB) only if you want a quiet always-on 8B-class machine that doubles as a desktop. At that price point, the value calculus tightens — the M4 Pro 24 GB at $1,399 starts looking close.
Skip the new $799 floor and chase a $499 eBay/Amazon residual instead, OR skip this tier entirely and step up to M4 Pro 24 GB ($1,399). At $799, the 16 GB ceiling for $200 less than the next tier up is a hard sell.

Watchouts

Apple discontinued the $599 base config (May 1, 2026); the 16 GB / 256 GB SKU is no longer listed on Apple's site. New floor is $799 (512 GB). Tim Cook flagged "several months" of supply constraint on the May 1 earnings call — DRAM shortage drove the cull.
$499 sale prices on Amazon are residual stock; eBay markups run $715–$979 on remaining 256 GB units. Verify storage tier before buying — a 256 GB unit fills fast with even a couple of 8B models.
16 GB is a hard ceiling. Soldered LPDDR5X. No upgrade path. If you grow into bigger models, you're selling and buying new.
Prefill latency hits harder at this tier. 120 GB/s is 1/5 of M4 Max. Long-context docs or coding work feels noticeably slower than higher Apple Silicon tiers.

Local vs cloud at this tier

● LOCAL WINS

Privacy, unlimited 8B chat / summarization / email drafting, zero-latency air-gapped runs. Useful as a general desktop if the local AI habit doesn't stick.

● CLOUD WINS

Anything bigger than 8B. ChatGPT Plus at $20/mo unlocks GPT-5 which is materially stronger than any 8B model for hard reasoning. Image-gen. Agents. Long-context docs.

At $799 (new floor) + ~$2/mo electricity, break-even vs a $100/mo ChatGPT Pro plan is ~8 months; vs $20/mo Plus, ~40+ months. The pre-discontinuation $499 unit broke even faster — if you can't find one, the 24 GB M4 Pro is probably the better next step.

Next step

Load this setup into the planner→