---
license: mit
task_categories:
  - text-generation
tags:
  - benchmark
  - inference
  - llm
  - nvidia
  - rtx-5090
  - llama-cpp
  - vllm
  - speed
  - quality
  - mmlu
  - gsm8k
  - humaneval
  - moe
size_categories:
  - n<1K
configs:
  - config_name: default
    data_files:
      - split: train
        path: benchmarks.csv
---

# RTX 5090 LLM Benchmarks

Speed and quality benchmarks for quantized LLMs on NVIDIA RTX 5090 32GB, measured with [llm-bench-rig](https://github.com/notwitcheer/llm-bench-rig).

## Quality Benchmarks

Generative evaluation through llama-server chat completions. Replicates standard benchmark methodology using custom evaluators — no `lm-evaluation-harness` dependency.

| Model | Params | Quant | MMLU | ARC-C | HellaSwag | GSM8K | HumanEval |
|-------|-------:|-------|-----:|------:|----------:|------:|----------:|
| Gemma 4 31B-it | 30.70B | Q6_K | 87.8 | 97.6 | 92.0 | 97.5 | 95.7 |
| Qwen3.6-27B | 26.90B | Q6_K | 87.9 | 96.9 | 95.4 | 97.3 | 18.9 |
| Qwen3.6-35B-A3B | 34.66B | UD-Q4_K_M | 85.0 | 95.7 | 93.4 | 96.7 | 37.2 |
| Qwen3-Coder-Next | 79.67B | UD-Q2_K_XL | 83.7 | 96.0 | 89.3 | 96.0 | 10.4 |
| Nemotron-Cascade-2 | 31.58B | Q4_K_M | 74.4 | 91.5 | 75.7 | 87.1 | 79.3 |
| gpt-oss-20b | 20.91B | Q4_K_M | 78.6 | 94.6 | 74.5 | 94.8 | 12.2 |

> All models benchmarked with 50% stratified sampling (seed=42), thinking disabled. Full per-model reports in [`reports/`](reports/).

### Methodology

| Benchmark | Dataset | Few-shot | Scoring | Items |
|-----------|---------|----------|---------|------:|
| MMLU | `cais/mmlu` | 5-shot | Letter extraction (A/B/C/D) | 14,042 |
| ARC-Challenge | `allenai/ai2_arc` | 25-shot | Letter extraction | 1,172 |
| HellaSwag | `Rowan/hellaswag` | 10-shot | Letter extraction | 10,042 |
| GSM8K | `openai/gsm8k` | 5-shot CoT | Exact numeric match | 1,319 |
| HumanEval | `openai/openai_humaneval` | 0-shot | pass@1 (code execution) | 164 |

All benchmarks run at `temperature=0` with `max_tokens=2048` (accommodates thinking/reasoning models). Multiple-choice tasks use generative letter extraction instead of loglikelihood scoring — scores are internally consistent for model comparison but may differ from logprob-based evaluations by 5-15%.

Full per-model reports with MMLU category breakdowns, parse reliability stats, and speed data: [`reports/`](reports/)

---

## Speed Benchmarks

### What's measured

- **Prompt processing (pp)**: parallel batched token throughput at context lengths 128, 512, 2048, 4096, 8192, 16384
- **Text generation (tg)**: sequential autoregressive token throughput at 128 tokens
- All models fully GPU-offloaded (ngl=99)

### Speed data schema

| Column | Description |
|--------|-------------|
| `model` | Model name |
| `architecture` | Dense or MoE (with active param count) |
| `params_b` | Total parameters in billions |
| `quant` | Quantization method |
| `size_gib` | File size in GiB |
| `engine` | Inference engine (llama.cpp or vLLM) |
| `backend` | Compute backend (CUDA) |
| `gpu` | GPU model |
| `vram_gb` | VRAM in GB |
| `test` | Benchmark test (pp128, pp512, ..., tg128) |
| `tokens_per_sec` | Throughput in tokens/second |
| `stddev` | Standard deviation |
| `date` | Benchmark date |

### Key findings

MoE (3B active) vs Dense (27B) on same-family Qwen3.6 models:
- Prompt processing: **2.4x faster** across all context lengths
- Text generation: **3.5x faster** (271 vs 77 t/s)
- Both degrade ~17% at 16K context (attention + VRAM, not parameter count)

---

## Hardware

| Component | Spec |
|-----------|------|
| GPU | NVIDIA GeForce RTX 5090 32GB (Blackwell, sm_120a) |
| CPU | AMD Ryzen 5 9600 (6c/12t) |
| RAM | 64GB DDR5-5600 |
| OS | Ubuntu 26.04 LTS |
| CUDA | 12.8 (patched for glibc 2.41) |

## Tooling

All benchmarks generated with [llm-bench-rig](https://github.com/notwitcheer/llm-bench-rig) — open-source pipeline for speed and quality benchmarks on GGUF and safetensors models.