Model benchmarks
llama3.1:8b local LLM performance
As of August 2026, llama3.1:8b runs at up to 107.7 tok/s for local inference (best of 7 community benchmark runs across 1 GPU).
OllamaQ4_K_M
Model size
8.0B
Peak speed
107.7 tok/s
Average speed
53.0 tok/s
Min memory
5.8 GB
Max context
65,536 tokens
Avg runtime / run
52s
Avg quality
44.9
Benchmark runs
7
GPUs tested
1
Quality by task
Average LLM-judged quality (0–100) with the run-to-run spread shown as a P5–P95 band, overall and for each benchmark task, across all 7 runs. The low and high columns show how much the judge’s score varies between runs, and need at least two runs to display.
| Task | P5 (low) | Avg | P95 (high) |
|---|---|---|---|
| Overall | 33.7 | 44.9 | 54.0 |
| Agent Workflow | 29.2 | 54.5 | 75.0 |
| Code Generation | 15.0 | 27.4 | 39.0 |
| Role Play & Narrative | 48.9 | 59.0 | 67.3 |
| Research & Analysis | 26.9 | 38.5 | 49.2 |
Performance by hardware and tool
Every hardware/tool/quantization combination llama3.1:8b has been benchmarked on, ranked by peak token generation speed. Last updated August 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| NVIDIA GeForce RTX 5080 | Ollama | Q4_K_M | 107.7 tok/s | 107.7 tok/s | 15.4 GB | 65,536 tokens | 43.4 | 1 |
| CPU only | Ollama | Q4_K_M | 74.0 tok/s | 43.9 tok/s | 5.8 GB | 8,192 tokens | 45.1 | 6 |
Frequently asked questions
- Is llama3.1:8b good for coding?
- In our benchmarks, llama3.1:8b scores 27.4/100 for coding. It runs at about 53.0 tok/s, so if you want more speed, Gemma4:E2B/QAT-MTP@131K is faster (~303.9 tok/s) and still scores well for coding (66.9/100).
- Is llama3.1:8b good for agentic (tool-using) tasks?
- In our benchmarks, llama3.1:8b scores 54.5/100 for agentic workflows. It runs at about 53.0 tok/s, so if you want more speed, Gemma4:E2B/QAT-MTP@131K is faster (~303.9 tok/s) and still scores well for agentic workflows (71.7/100).
- How fast is llama3.1:8b for local inference?
- Across 7 community benchmark runs, llama3.1:8b reaches up to 107.7 tok/s and averages 53.0 tok/s, with the fastest results on NVIDIA GeForce RTX 5080.
- How much memory does llama3.1:8b need?
- The leanest observed configuration used about 5.8 GB of memory (quantizations tested: Q4_K_M).
- Which tools have been used to run llama3.1:8b?
- Benchmarks were submitted using Ollama. Results are community-contributed and updated as new runs arrive.