llamaperf

Local model performance on your hardware

Find which open-weight LLMs fit in your GPU or Mac and compare the speeds people report on setups like yours.

What runs on your hardware?

Pick your GPU or Mac, then read the speeds people reported on it, or estimate which models fit and how fast they run.

Free to use. No account needed. Memory estimates and community measurements are labelled separately.

Looking for a particular model?

Search for a model to see its reported speeds across GPUs and Macs.

Local model performance reports from the community

These are individual setups, not a controlled benchmark. Compare GPU count, quantization, context and offloading before comparing speeds. How to read a report →

Model: Nemotron 3.5 Lightning
Compare setup details

Exact recorded values. Context may be a configured limit; matching filters does not establish identical prompts, offloading or concurrency.

Nemotron 3.5 Lightning 30B (3B active) Ternary

Unknown GPU · llama.cpp · 368,000 ctx

generation:
172.0 tokens/s
prompt processing (prefill):
7000.0 tokens/s
quant:
ternary 1.58-bit (GGUF)

User reports a ternary (1.58-bit) Nemotron 3.5 Lightning 30B-A3B GGUF running on a 12 GB GPU with 368K context, at 172 tok/s decode and 7000+ tok/s prefill. Setup is stock llama.cpp (validated on b11555+, b11796) and also the PrismML fork / Bonsai (validated on b10793), one file for both engines. The card does not name the specific GPU model.

Oct 26, 2026
Apple hardware
generation:
42.3 tokens/s
quant:
UD-Q3_K_XL (GGUF)

User reports Nemotron 3.5 Lightning 30B-A3B at 42.3 t/s on an Apple Mac mini M6 with 32 GB unified memory. Setup is LM Studio with its llama.cpp GGUF backend, UD-Q3_K_XL quant, and 32,768 tokens of loaded context. The 42.3 t/s is the median of three English-prose runs (Ruhr economy prompt), client-observed streaming decode throughput, not an overall mean or prefill rate.

Oct 13, 2026
Apple hardware
generation:
45.1 tokens/s
quant:
Q4_K_M (GGUF)

User reports Nemotron 3.5 Lightning 30B-A3B at 45.1 t/s on an Apple Mac mini M6 with 32 GB unified memory. Setup is LM Studio with its llama.cpp GGUF backend, GGUF Q4_K_M, and a loaded context of 32,768 tokens. The 45.1 t/s is the median of three English-prose runs and is client-observed streaming decode throughput, not an overall mean, peak, or prefill rate.

Oct 13, 2026

Nemotron 3.5 Lightning 30B (3B active)

Jetson AGX Orin 64GB · vLLM

generation:
80.0 tokens/s
agenticlong-context

User reports the current vLLM setup for Nemotron 3.5 Lightning 30B-A3B on a Jetson AGX Orin 64GB runs at around 80 tok/s while occupying roughly 62 GB of unified memory. The post is a benchmark plan: it proposes comparing SGLang and llama-server against the vLLM baseline, measuring cold and warm-prefix TTFT, decode tok/s, and resident memory across 4K to 128K prompts, a 32K shared-prefix four-agent workload, and constrained budgets of 32, 40 and 48 GB. No SGLang or llama-server results are given yet; the 80 tok/s figure is the existing vLLM baseline.

Oct 9, 2026
Tone: positiveNVIDIA hardware
generation:
110.0 tokens/s
quant:
Q4_0 (GGUF)
kv:
F16
codingagenticlong-context

User reports Nemotron-3.5-Lightning-30B-A3B at 110 tok/s writing code on two Tesla P100 16GB cards. Setup is llama.cpp b10970 with Q4_0 weights, F16 KV cache, tensor split across both cards, and the model's built-in MTP draft head at n-max 2. The same model reaches 95 tok/s on prose, 50 tok/s at 128k context, 39 tok/s at 256k, and 16 tok/s at 1M tokens. The study covers 545 speed measurements of 29 models from 2B to 122B parameters, all weights and KV cache in VRAM with no system RAM offload. A 119B MoE model runs 39 tok/s against 4.3 tok/s for a 70B dense model on the same cards. Tensor split makes dense models from 8B up 21-44% faster. A single P100 throttles to 906 MHz and loses 25% under sustained load, while two cards share the heat and lose 5.5%.

Oct 6, 2026
Tone: positiveNVIDIA hardware
generation:
72-87 tokens/s
quant:
Q4_K_M (GGUF)
agentictool-use

User reports Nemotron 3.5 Lightning 30B-A3B at 72 to 87 tok/s single-stream decode and about 2,600 tok/s prefill on a DGX Spark (GB10, 128GB unified memory). Setup is Ollama 0.32.9 with the Q4_K_M GGUF at the default 262,144 token context, 26GB resident at 100% GPU, with the built-in MTP speculative decoding active. Decode depends on the workload: 72 tok/s on prose and 84 to 87 tok/s on JSON and summaries. The same model served with vLLM using the NVFP4 checkpoint and DSpark draft model reached 108 tok/s decode and about 5,400 tok/s prefill. On one agent prompt the model answered in 485 tokens and 5.9s against 1,953 tokens and 26.0s for qwen3.5:35b-a3b.

Oct 4, 2026
Tone: positiveIntel hardware
generation:
91.9 tokens/s
prompt processing (prefill):
1760.0 tokens/s
quant:
Q4_0 (GGUF)
agenticcoding

User reports Nemotron 3.5 Lightning 30B-A3B at 91.91 tok/s decode on 2x Intel Arc Pro B60 24GB. Setup is llama.cpp SYCL with Q4_0 weights and an MTP Q8_0 drafter at --spec-draft-n-max 7, 22.18 GiB VRAM, 1,760 tok/s prefill at 12K context. MTP acceptance is 99.5-100% at every n-max; the model needs the whole card with no co-residence.

Oct 3, 2026
Tone: positiveNVIDIA hardware
generation:
80.0 tokens/s
prompt processing (prefill):
4737.5 tokens/s
quant:
NVFP4 (GGUF)
kv:
q8_0

User reports Nemotron 3.5 Lightning 30B-A3B at 79.98 t/s generation and 4737.46 t/s prompt processing on 2x RTX 5060 Ti 16GB. Setup is llama.cpp with NVFP4 GGUF weights and q8_0 KV cache, 1048576 token context, flash attention on, no MTP layers. The run processed a 54025-token prompt and generated 2156 tokens; the user notes it fits in 32 GB VRAM with no expert-layer offloading to RAM.

Sep 28, 2026
Tone: positiveNVIDIA hardware
quant:
W4A16

User compares a W4A16 quant of Nemotron 3.5 Lightning 30B-A3B against an IQ4_XS GGUF on an RTX 3090. Setup is vLLM for W4A16 and llama.cpp for IQ4_XS. The two are near-parity in instruction following benchmarks, with about 4.5x throughput by B16. The user describes the model as fast and reliable, and suitable for batch labelling and agentic responses.

Sep 7, 2026
Showing 1–9 of 9
Page 1 of 1

Community benchmarks snapshot

Records by GPU

AMD Strix Halo 128GB348NVIDIA RTX 3090271NVIDIA RTX 5090197NVIDIA DGX Spark99NVIDIA RTX 3060 12GB96NVIDIA RTX 5060 Ti 16GB88NVIDIA RTX Pro 6000 Blackwell81AMD Radeon AI PRO R9700 32GB81NVIDIA RTX 5070 Ti77NVIDIA RTX 409068

Records by model

3266 total
Qwen3.81567
Qwen3.6334
Gemma 4185
DeepSeek V4 Flash150
Qwen3.5145
DeepSeek V4.1 Flash103
other782

Records by engine

2308 total
llama.cpp1134
Strata234
vLLM227
MLX107
Ollama99
other507

Use cases

coding 937agentic 640long-context 442tool-use 281vision 174summarization 81math 77creative-writing 58multilingual 35rp 22text-generation 9reasoning 3
coding937agentic640long-context442tool-use281vision174summarization81math77creative-writing58

Median generation t/s by GPU

On Qwen3.8 27B at 4-bit, plain single-GPU runs. Full ranking

RTX 509080RTX Pro 600067RTX 5070 Ti55M5 Ultra 256GB50RTX 409040RX 7900 XTX40RTX 309038RX 7800 XT 16GB35RTX 5090 Laptop 24GB34RX 7900 XT32

Prompt processing, same model

M5 Ultra 256GB1800RTX 40901949RX 7900 XTX640RTX 3090561RX 7900 XT800M3 Ultra 512GB372V100 32GB941Radeon AI PRO R9700 32GB1287RTX 3080 20GB997Arc Pro B70512

Reports by model size

Qwen3.8 27B819Qwen3.8 125B · 6B active676Qwen3.6 35B · 3B active242DeepSeek V4 Flash 284B · 13B active133DeepSeek V4.1 Flash 552B · 16B active92Qwen3.6 27B81Gemma 4 26B · 4B active61GLM-5.3 320B · 18B active57

Quants

Q4_K_M395UD-Q4_K_XL189NVFP4174Q8_01194-bit106IQ4_XS86IQ3_XXS85IQ3_S84Q4_K_XL83Q480