Is Ornith-1.5-35B-A3B-GSQ-RCO-3bit good for agentic (tool-using) tasks?

In our benchmarks, Ornith-1.5-35B-A3B-GSQ-RCO-3bit scores 84.5/100 for agentic workflows. It runs at about 97.0 tok/s, so if you want more speed, cyber-tiel-coder-35b-a3b-mtp is faster (~115.9 tok/s) and still scores well for agentic workflows (86.4/100).

How fast is Ornith-1.5-35B-A3B-GSQ-RCO-3bit for local inference?

Across 6 community benchmark runs, Ornith-1.5-35B-A3B-GSQ-RCO-3bit reaches up to 113.4 tok/s and averages 97.0 tok/s, with the fastest results on AMD Radeon RX 7800 XT.

How much memory does Ornith-1.5-35B-A3B-GSQ-RCO-3bit need?

The leanest observed configuration used about 14.2 GB of memory.

Which tools have been used to run Ornith-1.5-35B-A3B-GSQ-RCO-3bit?

Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.

Model benchmarks

Ornith-1.5-35B-A3B-GSQ-RCO-3bit local LLM performance

As of October 2026, Ornith-1.5-35B-A3B-GSQ-RCO-3bit runs at up to 113.4 tok/s for local inference (best of 6 community benchmark runs across 2 GPUs).

llama.cpp

ShareReddit X

Model size

35B

Peak speed

113.4 tok/s

Average speed

97.0 tok/s

Avg PP

1091.3 tok/s

Min memory

14.2 GB

Max context

131,072 tokens

Avg output / run

28,968 tokens

Avg runtime / run

5m 13s

Avg quality

83.6

Benchmark runs

GPUs tested

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 6 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	80.8	83.6	85.4
Agent Workflow	77.7	84.5	90.2
Code Generation	70.9	77.4	83.4
Role Play & Narrative	82.2	87.0	90.6
Research & Analysis	82.7	85.5	88.2

Performance by hardware and tool

Every hardware/tool/quantization combination Ornith-1.5-35B-A3B-GSQ-RCO-3bit 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 7800 XT	llama.cpp	—	113.4 tok/s	102.4 tok/s	14.2 GB	131,072 tokens	83.8	2
NVIDIA GeForce RTX 5060	llama.cpp	—	104.3 tok/s	94.2 tok/s	16.7 GB	131,072 tokens	83.5	4

Benchmark runs

All 6 Ornith-1.5-35B-A3B-GSQ-RCO-3bit runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.

Open run page →

Frequently asked questions

Is Ornith-1.5-35B-A3B-GSQ-RCO-3bit good for agentic (tool-using) tasks?: In our benchmarks, Ornith-1.5-35B-A3B-GSQ-RCO-3bit scores 84.5/100 for agentic workflows. It runs at about 97.0 tok/s, so if you want more speed, cyber-tiel-coder-35b-a3b-mtp is faster (~115.9 tok/s) and still scores well for agentic workflows (86.4/100).
How fast is Ornith-1.5-35B-A3B-GSQ-RCO-3bit for local inference?: Across 6 community benchmark runs, Ornith-1.5-35B-A3B-GSQ-RCO-3bit reaches up to 113.4 tok/s and averages 97.0 tok/s, with the fastest results on AMD Radeon RX 7800 XT.
How much memory does Ornith-1.5-35B-A3B-GSQ-RCO-3bit need?: The leanest observed configuration used about 14.2 GB of memory.
Which tools have been used to run Ornith-1.5-35B-A3B-GSQ-RCO-3bit?: Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.