Model benchmarks
swift-1.5-iq2_xs local LLM performance
As of November 2026, swift-1.5-iq2_xs runs at up to 91.4 tok/s for local inference (best of 8 community benchmark runs across 4 GPUs).
Model size
Unknown
Peak speed
91.4 tok/s
Average speed
67.0 tok/s
Avg PP
377.0 tok/s
Min memory
13.0 GB
Max context
65,536 tokens
Avg output / run
35,329 tokens
Avg runtime / run
11m
Avg quality
80.9
Benchmark runs
8
GPUs tested
4
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 8 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 | 60.1 | 80.9 | 87.1 |
| Agent Workflow | 40.1 | 78.5 | 91.5 |
| Code Generation | 26.3 | 69.1 | 83.5 |
| Role Play & Narrative | 88.2 | 91.5 | 94.3 |
| Research & Analysis | 80.1 | 84.4 | 88.9 |
Performance by hardware and tool
Every hardware/tool/quantization combination swift-1.5-iq2_xs has been benchmarked on, ranked by peak token generation speed. Last updated November 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| AMD Radeon RX 7900 XTX | Strata | — | 91.4 tok/s | 80.2 tok/s | 55.9 GB | 65,536 tokens | 77.7 | 5 |
| NVIDIA RTX 5000 Ada Generation Laptop GPU | Strata | — | 66.4 tok/s | 66.4 tok/s | 52.6 GB | 65,536 tokens | 85.2 | 1 |
| NVIDIA GeForce RTX 3070 Ti | polystrata | — | 43.4 tok/s | 43.4 tok/s | 41.3 GB | 65,536 tokens | 86.6 | 1 |
| AMD Radeon RX 9070 XT | llama.cpp | — | 25.2 tok/s | 25.2 tok/s | 13.0 GB | 65,536 tokens | 86.8 | 1 |
Benchmark runs
All 8 swift-1.5-iq2_xs runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is swift-1.5-iq2_xs good for coding?
- In our benchmarks, swift-1.5-iq2_xs scores 69.1/100 for coding. It runs at about 67.0 tok/s, so if you want more speed, qwen3.8-27b-Swift1.5-Uncensored-ATX-AI3 is faster (~85.5 tok/s) and still scores well for coding (84.8/100).
- Is swift-1.5-iq2_xs good for agentic (tool-using) tasks?
- In our benchmarks, swift-1.5-iq2_xs scores 78.5/100 for agentic workflows. It runs at about 67.0 tok/s, so if you want more speed, Nex-N2.5-mini-APEX-Mini is faster (~88.2 tok/s) and still scores well for agentic workflows (91.3/100).
- How fast is swift-1.5-iq2_xs for local inference?
- Across 8 community benchmark runs, swift-1.5-iq2_xs reaches up to 91.4 tok/s and averages 67.0 tok/s, with the fastest results on AMD Radeon RX 7900 XTX.
- How much memory does swift-1.5-iq2_xs need?
- The leanest observed configuration used about 13.0 GB of memory.
- Which tools have been used to run swift-1.5-iq2_xs?
- Benchmarks were submitted using Strata, llama.cpp, polystrata. Results are community-contributed and updated as new runs arrive.