Model benchmarks

Ornith-1.5-9B-MTP-Q4_0_ROCMFP4_COHERENT local LLM performance

As of August 2026, Ornith-1.5-9B-MTP-Q4_0_ROCMFP4_COHERENT runs at up to 59.3 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).

LM StudioQ4_0
ShareRedditX

Model size

9B

Peak speed

59.3 tok/s

Average speed

58.3 tok/s

Min memory

4.4 GB

Max context

8.192 tokens

Avg runtime / run

4m 47s

Avg quality

57.7

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.

TaskP5 (low)AvgP95 (high)
Overall57.157.758.4
Agent Workflow86.188.290.2
Code Generation6.09.312.6
Role Play & Narrative67.271.375.4
Research & Analysis50.262.274.2

Performance by hardware and tool

Every hardware/tool/quantization combination Ornith-1.5-9B-MTP-Q4_0_ROCMFP4_COHERENT has been benchmarked on, ranked by peak token generation speed. Last updated August 2026.

HardwareToolQuantBest tok/sAvg tok/sMemoryContextQualityRuns
AMD Radeon RX 6700/6700 XT/6750 XT / 6800M/6850M XTLM StudioQ4_059.3 tok/s58.3 tok/s4.4 GB8.192 tokens57.72

Benchmark runs

All 2 Ornith-1.5-9B-MTP-Q4_0_ROCMFP4_COHERENT runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.

Frequently asked questions

Is Ornith-1.5-9B-MTP-Q4_0_ROCMFP4_COHERENT good for coding?
In our benchmarks, Ornith-1.5-9B-MTP-Q4_0_ROCMFP4_COHERENT scores 9.3/100 for coding. It runs at about 58.3 tok/s, so if you want more speed, unsloth/Qwen3.8-27B-GGUF:IQ3_S is faster (~104.3 tok/s) and still scores well for coding (83.0/100).
Is Ornith-1.5-9B-MTP-Q4_0_ROCMFP4_COHERENT good for agentic (tool-using) tasks?
In our benchmarks, Ornith-1.5-9B-MTP-Q4_0_ROCMFP4_COHERENT scores 88.2/100 for agentic workflows. It runs at about 58.3 tok/s, so if you want more speed, granite-4.2-8b-Q4_K_L is faster (~71.6 tok/s) and still scores well for agentic workflows (92.9/100).
How fast is Ornith-1.5-9B-MTP-Q4_0_ROCMFP4_COHERENT for local inference?
Across 2 community benchmark runs, Ornith-1.5-9B-MTP-Q4_0_ROCMFP4_COHERENT reaches up to 59.3 tok/s and averages 58.3 tok/s, with the fastest results on AMD Radeon RX 6700/6700 XT/6750 XT / 6800M/6850M XT.
How much memory does Ornith-1.5-9B-MTP-Q4_0_ROCMFP4_COHERENT need?
The leanest observed configuration used about 4.4 GB of memory (quantizations tested: Q4_0).
Which tools have been used to run Ornith-1.5-9B-MTP-Q4_0_ROCMFP4_COHERENT?
Benchmarks were submitted using LM Studio. Results are community-contributed and updated as new runs arrive.