Model benchmarks
mistralai/ministral-3-3b local LLM performance
As of October 2026, mistralai/ministral-3-3b runs at up to 12.7 tok/s for local inference (best of 3 community benchmark runs across 3 GPUs).
Model size
3B
Peak speed
12.7 tok/s
Average speed
6.5 tok/s
Avg PP
275.0 tok/s
Min memory
6.7 GB
Max context
65,536 tokens
Avg output / run
8,342 tokens
Avg runtime / run
30m 13s
Avg quality
45.7
Benchmark runs
3
GPUs tested
3
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 3 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 | 41.1 | 45.7 | 49.3 |
| Agent Workflow | 41.0 | 54.0 | 62.1 |
| Code Generation | 17.4 | 25.7 | 32.7 |
| Role Play & Narrative | 50.8 | 63.4 | 78.7 |
| Research & Analysis | 35.3 | 39.8 | 42.9 |
Performance by hardware and tool
Every hardware/tool/quantization combination mistralai/ministral-3-3b has been benchmarked on, ranked by peak token generation speed. Last updated October 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| AMD Radeon Vega Series / Radeon Vega Mobile Series | Bionic | — | 12.7 tok/s | 12.7 tok/s | 14.3 GB | 8,192 tokens | 40.4 | 1 |
| NVIDIA GeForce RTX 3080 Laptop GPU | LM Studio | — | 3.7 tok/s | 3.7 tok/s | n/a | 65,536 tokens | 47.3 | 1 |
| NVIDIA GeForce GTX 1050 | LM Studio | Q4_K_M | 3.2 tok/s | 3.2 tok/s | 6.7 GB | 8,192 tokens | 49.5 | 1 |
Benchmark runs
All 3 mistralai/ministral-3-3b runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is mistralai/ministral-3-3b good for coding?
- In our benchmarks, mistralai/ministral-3-3b scores 25.7/100 for coding. It runs at about 6.5 tok/s, so if you want more speed, Qwen3.8-27B-oQ4e-fp16-mtp is faster (~38.6 tok/s) and still scores well for coding (82.6/100).
- Is mistralai/ministral-3-3b good for agentic (tool-using) tasks?
- In our benchmarks, mistralai/ministral-3-3b scores 54.0/100 for agentic workflows. It runs at about 6.5 tok/s, so if you want more speed, muse-glimmer:latest is faster (~33.3 tok/s) and still scores well for agentic workflows (91.6/100).
- How fast is mistralai/ministral-3-3b for local inference?
- Across 3 community benchmark runs, mistralai/ministral-3-3b reaches up to 12.7 tok/s and averages 6.5 tok/s, with the fastest results on AMD Radeon Vega Series / Radeon Vega Mobile Series.
- How much memory does mistralai/ministral-3-3b need?
- The leanest observed configuration used about 6.7 GB of memory (quantizations tested: Q4_K_M).
- Which tools have been used to run mistralai/ministral-3-3b?
- Benchmarks were submitted using Bionic, LM Studio. Results are community-contributed and updated as new runs arrive.