Best Local LLMs for 96GB RAM
Ranked open-weight models that run well on a 96GB machine, with estimated speed and the exact ollama command for each.
Try: Qwen3 8B, RTX 4090, MacBook Pro M4, 16GB, iPhone 17 Pro
Best local models for ~96GB
12 picksEstimates assume a representative Apple-Silicon machine with 96GB unified memory. Tok/s are ModelFit estimates, not measured benchmarks; typical error is the median gap to measured runs on hardware the model was not fitted on (how the estimates work). Run the wizard for figures tuned to your exact chip.
- 01
Kolibri 1
Best for reasoning, german, long context. Strong fit for 96 GB RAM with balanced speed and quality.
$ollama run kolibri-1-q4- Fit
- Runs wellB · 35
- Best for
- Reasoning, German, Long context
- Perf
- est. 112 tok/s, typical error 10% · first token ~0.5s
- 02
Kolibri 1 (IQ4_XS)
Best for reasoning, german, long context. Strong fit for 96 GB RAM with balanced speed and quality.
$ollama run kolibri-1-q4xs- Fit
- Runs wellB · 43
- Best for
- Reasoning, German, Long context
- Perf
- est. 121 tok/s, typical error 10% · first token ~0.5s
- 03
Qwen3.6 35B-A3B (Q8)
Best for reasoning, coding, agents. Strong fit for 96 GB RAM with balanced speed and quality.
$ollama run qwen3.6:35b-a3b-q8_0- Fit
- Runs wellB · 50
- Best for
- Reasoning, Coding, Agents
- Perf
- est. 99 tok/s, typical error 10% · first token ~0.4s
- 04
Nemotron 3 Super 120B-A12B (IQ4_XS)
This model may feel memory-heavy on 96 GB RAM, but it is still listed for balanced speed and quality.
$ollama run nemotron-3-super-120b-q4xs- Fit
- Runs wellC · 16
- Best for
- Agentic workflows, Long context, Reasoning
- Perf
- est. 48 tok/s, typical error 10% · first token ~1.7s
- 05
Qwen3.5 122B-A10B Instruct
This model may feel memory-heavy on 96 GB RAM, but it is still listed for balanced speed and quality.
$ollama run qwen3.5:122b-a10b- Fit
- Runs wellC · 6
- Best for
- Frontier-level reasoning, Complex tasks
- Perf
- est. 62 tok/s, typical error 10% · first token ~1.1s
- 06
GPT-OSS 120B
This model may feel memory-heavy on 96 GB RAM, but it is still listed for balanced speed and quality.
$ollama run gpt-oss:120b- Fit
- Runs wellC · 15
- Best for
- Reasoning, Coding, Agents
- Perf
- est. 90 tok/s, typical error 10% · first token ~0.7s
- 07
Qwen3-Next 80B-A3B
Best for chat, coding, long context. Strong fit for 96 GB RAM with balanced speed and quality.
$ollama run qwen3-next:80b- Fit
- Runs wellB · 34
- Best for
- Chat, Coding, Long Context
- Perf
- est. 124 tok/s, typical error 10% · first token ~0.4s
- 08
Ling-3.0-flash (IQ4_XS)
This model may feel memory-heavy on 96 GB RAM, but it is still listed for balanced speed and quality.
$ollama run ling-3.0-flash-q4xs- Fit
- Runs wellC · 11
- Best for
- Agentic coding, Long context, Reasoning
- Perf
- est. 102 tok/s, typical error 10% · first token ~0.7s
- 09
Nemotron 3.5 Lightning 30B-A3B
Best for agentic, coding, long context. Strong fit for 96 GB RAM with balanced speed and quality.
$ollama run nemotron-3.5-lightning:30b- Fit
- Lots of roomA · 67
- Best for
- Agentic, Coding, Long context
- Perf
- est. 99 tok/s, typical error 10% · first token ~0.5s
- 10
Qwen3.5 35B-A3B Instruct (Q8)
Best for reasoning, coding, agent scenarios. Strong fit for 96 GB RAM with balanced speed and quality.
$ollama run qwen3.5:35b-a3b-q8_0- Fit
- Runs wellB · 50
- Best for
- Reasoning, Coding, Agent scenarios
- Perf
- est. 99 tok/s, typical error 10% · first token ~0.4s
- 11
Gemma 4 26B-A4B (Q8)
Best for chat, coding, multimodal. Strong fit for 96 GB RAM with balanced speed and quality.
$ollama run gemma4:26b-a4b-it-q8_0- Fit
- Lots of roomA · 63
- Best for
- Chat, Coding, Multimodal
- Perf
- est. 76 tok/s, typical error 10% · first token ~0.5s
- 12
Llama 4 Scout
This model may feel memory-heavy on 96 GB RAM, but it is still listed for balanced speed and quality.
$ollama run llama4:scout- Fit
- Runs wellC · 13
- Best for
- Long context, Quality, Multimodal
- Perf
- est. 45 tok/s, typical error 10% · first token ~1.6s
Get told when a better model fits your MacBook Air M5 16 GB
Get one email when a new open-weight model is a better fit for this machine.
For: MacBook Air M5 16 GB
You'll also get the weekly ModelFit email. By signing up you agree to our Privacy Policy. Unsubscribe from either email anytime, in one click.
Browse by RAM tier
Want an exact recommendation?
The wizard tunes picks and speed estimates to your exact device, chip, and RAM.