Model benchmarks

Inkling-2-Mini-oQ8e local LLM performance

As of November 2026, Inkling-2-Mini-oQ8e runs at up to 58.7 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).

oMLXoQ8e
ShareRedditX

Model size

1.6B

Peak speed

58.7 tok/s

Average speed

57.7 tok/s

Avg PP

930.9 tok/s

Min memory

18.0 GB

Max context

262,144 tokens

Avg output / run

34,634 tokens

Avg runtime / run

10m 6s

Avg quality

73.9

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)
Overall66.873.981.0
Agent Workflow20.853.285.5
Code Generation80.783.786.7
Role Play & Narrative73.674.876.0
Research & Analysis83.884.084.1

Performance by hardware and tool

Every hardware/tool/quantization combination Inkling-2-Mini-oQ8e has been benchmarked on, ranked by peak token generation speed. Last updated November 2026.

HardwareToolQuantBest tok/sAvg tok/sMemoryContextQualityRuns
Apple M5 MaxoMLXoQ8e58.7 tok/s57.7 tok/s18.0 GB262,144 tokens73.92

Benchmark runs

All 2 Inkling-2-Mini-oQ8e runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.

Frequently asked questions

Is Inkling-2-Mini-oQ8e good for coding?
In our benchmarks, Inkling-2-Mini-oQ8e scores 83.7/100 for coding. It runs at about 57.7 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 Inkling-2-Mini-oQ8e good for agentic (tool-using) tasks?
In our benchmarks, Inkling-2-Mini-oQ8e scores 53.2/100 for agentic workflows. It runs at about 57.7 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 Inkling-2-Mini-oQ8e for local inference?
Across 2 community benchmark runs, Inkling-2-Mini-oQ8e reaches up to 58.7 tok/s and averages 57.7 tok/s, with the fastest results on Apple M5 Max.
How much memory does Inkling-2-Mini-oQ8e need?
The leanest observed configuration used about 18.0 GB of memory (quantizations tested: oQ8e).
Which tools have been used to run Inkling-2-Mini-oQ8e?
Benchmarks were submitted using oMLX. Results are community-contributed and updated as new runs arrive.