Model benchmarks
Ornith-1.5-35B-A3B.APEX-I-NanoPlus local LLM performance
As of October 2026, Ornith-1.5-35B-A3B.APEX-I-NanoPlus runs at up to 87.2 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
Model size
35B
Peak speed
87.2 tok/s
Average speed
86.5 tok/s
Avg PP
1793.2 tok/s
Min memory
13.8 GB
Max context
65,536 tokens
Avg output / run
41,724 tokens
Avg runtime / run
8m 37s
Avg quality
72.5
Benchmark runs
2
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 2 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 | 66.9 | 72.5 | 78.0 |
| Agent Workflow | 35.6 | 60.6 | 85.6 |
| Code Generation | 80.1 | 80.9 | 81.7 |
| Role Play & Narrative | 58.3 | 61.5 | 64.7 |
| Research & Analysis | 86.4 | 86.8 | 87.2 |
Performance by hardware and tool
Every hardware/tool/quantization combination Ornith-1.5-35B-A3B.APEX-I-NanoPlus 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 RX 9070/9070 XT/9070 GRE | llama.cpp | — | 87.2 tok/s | 86.5 tok/s | 13.8 GB | 65,536 tokens | 72.5 | 2 |
Benchmark runs
All 2 Ornith-1.5-35B-A3B.APEX-I-NanoPlus runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is Ornith-1.5-35B-A3B.APEX-I-NanoPlus good for coding?
- In our benchmarks, Ornith-1.5-35B-A3B.APEX-I-NanoPlus scores 80.9/100 for coding. It runs at about 86.5 tok/s, so if you want more speed, Tiel-Coder-35B-A3B-MLX-oQ4e is faster (~107.9 tok/s) and still scores well for coding (81.5/100).
- Is Ornith-1.5-35B-A3B.APEX-I-NanoPlus good for agentic (tool-using) tasks?
- In our benchmarks, Ornith-1.5-35B-A3B.APEX-I-NanoPlus scores 60.6/100 for agentic workflows. It runs at about 86.5 tok/s, so if you want more speed, qwen3.8-flash-next-iq3_s is faster (~116.8 tok/s) and still scores well for agentic workflows (88.4/100).
- How fast is Ornith-1.5-35B-A3B.APEX-I-NanoPlus for local inference?
- Across 2 community benchmark runs, Ornith-1.5-35B-A3B.APEX-I-NanoPlus reaches up to 87.2 tok/s and averages 86.5 tok/s, with the fastest results on AMD Radeon RX 9070/9070 XT/9070 GRE.
- How much memory does Ornith-1.5-35B-A3B.APEX-I-NanoPlus need?
- The leanest observed configuration used about 13.8 GB of memory.
- Which tools have been used to run Ornith-1.5-35B-A3B.APEX-I-NanoPlus?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.