Best local LLMs for the Apple M5 Ultra (96 GB)

Each list assumes the context that task needs, so a model that fits for chat may rank lower for long documents.

General chat at 8K context

  1. 91Excellent quality (A+) · 91 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 100/100 (quality, fit, speed and memory use) sets the ranking. Nemotron 3.5 Bolt 74B-A9B Q6_K · 72.3 tok/s · Runs well (tight fit)
  2. 91Excellent quality (A+) · 91 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 100/100 (quality, fit, speed and memory use) sets the ranking. Sage-A1.5 35B-A3B Q8_0 · 118 tok/s · Runs great
  3. 90Excellent quality (A+) · 90 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. Ornith 1.5 35B-A3B Q8_0 · 118 tok/s · Runs great
  4. 90Excellent quality (A+) · 90 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. Kolibri 1 (78B-A3.5B) Q6_K · 133 tok/s · Runs well (tight fit)
  5. 89Excellent quality (A+) · 89 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. Apodex 1.1 Mini (35B-A3B) Q8_0 · 118 tok/s · Runs great

Coding at 32K context

  1. 91Excellent quality (A+) · 91 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 100/100 (quality, fit, speed and memory use) sets the ranking. Nemotron 3.5 Bolt 74B-A9B Q6_K · 72.3 tok/s · Runs well (tight fit)
  2. 91Excellent quality (A+) · 91 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. KAT-Coder V2.5 Dev (35B-A3B) Q8_0 · 118 tok/s · Runs great
  3. 90Excellent quality (A+) · 90 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. Kolibri 1 (78B-A3.5B) Q6_K · 133 tok/s · Runs well (tight fit)
  4. 89Excellent quality (A+) · 89 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. Apodex 1.1 Mini (35B-A3B) Q8_0 · 118 tok/s · Runs great
  5. 87Excellent quality (A+) · 87 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 98/100 (quality, fit, speed and memory use) sets the ranking. Qwen AgentWorld 35B-A3B Q8_0 · 118 tok/s · Runs great

Reasoning at 32K context

  1. 91Excellent quality (A+) · 91 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 100/100 (quality, fit, speed and memory use) sets the ranking. Nemotron 3.5 Bolt 74B-A9B Q6_K · 72.3 tok/s · Runs well (tight fit)
  2. 91Excellent quality (A+) · 91 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. KAT-Coder V2.5 Dev (35B-A3B) Q8_0 · 118 tok/s · Runs great
  3. 90Excellent quality (A+) · 90 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. Kolibri 1 (78B-A3.5B) Q6_K · 133 tok/s · Runs well (tight fit)
  4. 89Excellent quality (A+) · 89 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. Apodex 1.1 Mini (35B-A3B) Q8_0 · 118 tok/s · Runs great
  5. 87Excellent quality (A+) · 87 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 98/100 (quality, fit, speed and memory use) sets the ranking. Agents-A1 35B-A3B Q8_0 · 118 tok/s · Runs great

Documents (RAG) at 32K context

  1. 90Excellent quality (A+) · 90 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. Ornith 1.5 35B-A3B Q8_0 · 118 tok/s · Runs great
  2. 90Excellent quality (A+) · 90 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. Kolibri 1 (78B-A3.5B) Q6_K · 133 tok/s · Runs well (tight fit)
  3. 89Excellent quality (A+) · 89 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. Apodex 1.1 Mini (35B-A3B) Q8_0 · 118 tok/s · Runs great
  4. 89Excellent quality (A+) · 89 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. Ornith 1.0 35B-A3B Q8_0 · 118 tok/s · Runs great
  5. 92Excellent quality (A+) · 92 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 98/100 (quality, fit, speed and memory use) sets the ranking. Qwen3.6 35B-A3B Q6_K · 143 tok/s · Runs great

Agents & tools at 64K context

  1. 91Excellent quality (A+) · 91 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 100/100 (quality, fit, speed and memory use) sets the ranking. KAT-Coder V2.5 Dev (35B-A3B) Q8_0 · 118 tok/s · Runs great
  2. 91Excellent quality (A+) · 91 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 100/100 (quality, fit, speed and memory use) sets the ranking. Nemotron 3.5 Bolt 74B-A9B Q6_K · 72.3 tok/s · Runs well (tight fit)
  3. 89Excellent quality (A+) · 89 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. Apodex 1.1 Mini (35B-A3B) Q8_0 · 118 tok/s · Runs great
  4. 90Excellent quality (A+) · 90 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. Kolibri 1 (78B-A3.5B) Q6_K · 133 tok/s · Runs well (tight fit)
  5. 87Excellent quality (A+) · 87 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 98/100 (quality, fit, speed and memory use) sets the ranking. Agents-A1 35B-A3B Q8_0 · 118 tok/s · Runs great

Reverse engineering at 32K context

  1. 92Excellent quality (A+) · 92 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 98/100 (quality, fit, speed and memory use) sets the ranking. Qwen3.5 35B-A3B Q6_K · 143 tok/s · Runs great
  2. 88Excellent quality (A+) · 88 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 97/100 (quality, fit, speed and memory use) sets the ranking. GLM-4.7 Flash (30B-A3B) Q8_0 · 118 tok/s · Runs great
  3. 92Excellent quality (A+) · 92 of 100: how capable the model is at F16, whatever it runs on. Match score 96/100 (quality, fit, speed and memory use) sets the ranking. Gemma 4 26B-A4B F16 · 58.3 tok/s · Runs great
  4. 93Excellent quality (A+) · 93 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 96/100 (quality, fit, speed and memory use) sets the ranking. Qwen3.6 35B-A3B Q8_0 · 118 tok/s · Runs great
  5. 93Excellent quality (A+) · 93 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 96/100 (quality, fit, speed and memory use) sets the ranking. MiniCPM-V 4.7 35B-A3B Q8_0 · 118 tok/s · Runs great

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.

Was this page helpful?