Model benchmarks
Ornith-1.5-35B-A3B-oQ8e-mtp local LLM performance
As of October 2026, Ornith-1.5-35B-A3B-oQ8e-mtp runs at up to 129.1 tok/s for local inference (best of 11 community benchmark runs across 1 GPU).
Model size
35B
Peak speed
129.1 tok/s
Average speed
104.4 tok/s
Avg PP
3656.2 tok/s
Min memory
36.5 GB
Max context
262,144 tokens
Avg output / run
40,144 tokens
Avg runtime / run
6m 59s
Avg quality
84.3
Benchmark runs
11
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 11 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 | 81.4 | 84.3 | 85.5 |
| Agent Workflow | 72.4 | 84.8 | 90.8 |
| Code Generation | 74.9 | 78.9 | 83.9 |
| Role Play & Narrative | 83.6 | 87.4 | 91.8 |
| Research & Analysis | 82.1 | 85.9 | 89.5 |
Performance by hardware and tool
Every hardware/tool/quantization combination Ornith-1.5-35B-A3B-oQ8e-mtp 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 |
|---|---|---|---|---|---|---|---|---|
| Apple M5 Max | oMLX | oQ8e | 129.1 tok/s | 116.6 tok/s | 36.7 GB | 262,144 tokens | 84.8 | 5 |
| Apple M5 Max | oMLX | — | 98.3 tok/s | 94.2 tok/s | 36.5 GB | 262,144 tokens | 83.8 | 6 |
Benchmark runs
All 11 Ornith-1.5-35B-A3B-oQ8e-mtp runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is Ornith-1.5-35B-A3B-oQ8e-mtp good for coding?
- In our benchmarks, Ornith-1.5-35B-A3B-oQ8e-mtp scores 78.9/100 for coding. It runs at about 104.4 tok/s, so if you want more speed, turbo is faster (~124.1 tok/s) and still scores well for coding (85.3/100).
- Is Ornith-1.5-35B-A3B-oQ8e-mtp good for agentic (tool-using) tasks?
- In our benchmarks, Ornith-1.5-35B-A3B-oQ8e-mtp scores 84.8/100 for agentic workflows. It runs at about 104.4 tok/s, so if you want more speed, Qwen3.8-27B-IQ5R-V4-DFlash2-TP2-i8KV-r21 is faster (~133.3 tok/s) and still scores well for agentic workflows (91.8/100).
- How fast is Ornith-1.5-35B-A3B-oQ8e-mtp for local inference?
- Across 11 community benchmark runs, Ornith-1.5-35B-A3B-oQ8e-mtp reaches up to 129.1 tok/s and averages 104.4 tok/s, with the fastest results on Apple M5 Max.
- How much memory does Ornith-1.5-35B-A3B-oQ8e-mtp need?
- The leanest observed configuration used about 36.5 GB of memory (quantizations tested: oQ8e).
- Which tools have been used to run Ornith-1.5-35B-A3B-oQ8e-mtp?
- Benchmarks were submitted using oMLX. Results are community-contributed and updated as new runs arrive.