Model benchmarks
ornith-1.5:35b-a3b local LLM performance
As of October 2026, ornith-1.5:35b-a3b runs at up to 95.8 tok/s for local inference (best of 5 community benchmark runs across 3 GPUs).
Model size
35B
Peak speed
95.8 tok/s
Average speed
60.6 tok/s
Avg PP
614.6 tok/s
Min memory
19.1 GB
Max context
262,144 tokens
Avg output / run
19,193 tokens
Avg runtime / run
13m 28s
Avg quality
80.3
Benchmark runs
5
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 5 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 | 71.3 | 80.3 | 85.2 |
| Agent Workflow | 76.0 | 81.4 | 88.2 |
| Code Generation | 59.3 | 73.9 | 82.7 |
| Role Play & Narrative | 70.8 | 82.3 | 90.4 |
| Research & Analysis | 78.9 | 83.7 | 88.4 |
Performance by hardware and tool
Every hardware/tool/quantization combination ornith-1.5:35b-a3b 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 9060 XT | llama.cpp | — | 95.8 tok/s | 95.6 tok/s | 21.4 GB | 262,144 tokens | 73.4 | 2 |
| Apple M4 Pro | oMLX | — | 71.1 tok/s | 71.1 tok/s | 19.1 GB | 140,000 tokens | 84.9 | 1 |
| AMD Radeon RX 7600 XT | LM Studio | — | 23.6 tok/s | 23.6 tok/s | n/a | 32,768 tokens | 84.7 | 1 |
| CPU only | LM Studio | Q4_K_M | 16.8 tok/s | 16.8 tok/s | n/a | 65,536 tokens | 85.3 | 1 |
Benchmark runs
All 5 ornith-1.5:35b-a3b runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is ornith-1.5-35b-a3b good for coding?
- In our benchmarks, ornith-1.5-35b-a3b scores 73.9/100 for coding. It runs at about 60.6 tok/s, so if you want more speed, ornith-1.5-35b-a3b-apex-mtp-i-quality is faster (~105.2 tok/s) and still scores well for coding (83.6/100).
- Is ornith-1.5-35b-a3b good for agentic (tool-using) tasks?
- In our benchmarks, ornith-1.5-35b-a3b scores 81.4/100 for agentic workflows. It runs at about 60.6 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 ornith-1.5:35b-a3b for local inference?
- Across 5 community benchmark runs, ornith-1.5:35b-a3b reaches up to 95.8 tok/s and averages 60.6 tok/s, with the fastest results on AMD Radeon RX 9060 XT.
- How much memory does ornith-1.5:35b-a3b need?
- The leanest observed configuration used about 19.1 GB of memory (quantizations tested: Q4_K_M).
- Which tools have been used to run ornith-1.5:35b-a3b?
- Benchmarks were submitted using LM Studio, llama.cpp, oMLX. Results are community-contributed and updated as new runs arrive.