Nex-N2.5 mini · MLX oQ2
A community mixed-precision conversion by Vontra for Apple Silicon.
Model
Converted from the BF16 Nex-N2.5-mini checkpoint using oMLX oQ2. The oQ label is a target, not a claim that every tensor uses the same precision; see the per-module quantisation entries in config.json. This text-and-vision model uses the qwen3_5_moe architecture. The inspected BF16 checkpoint contained no matching MTP tensors, despite its configuration declaring one MTP layer. This release does not provide tested MTP decoding; keep MTP disabled.
Download and use
hf download Vontra/Nex-N2.5-mini-MLX-oQ2 --local-dir ./Nex-N2.5-mini-MLX-oQ2
Add the folder to oMLX model directories, refresh the model list and select it. Basic inference was tested with oMLX 0.6.4.
Use the upstream-recommended sampling:
{
"temperature": 0.7,
"top_p": 0.95,
"top_k": 40
}
Set these explicitly in your client or model settings. Our sampled oQ2 retests also used reasoning_effort="none" and max_tokens=4096; these are test conditions, not an upstream recommendation to disable reasoning. Avoid greedy decoding for oQ2, which reproduced a repetition loop in our tests.
Validation and limitations
Tested on 9 September 2026 through oMLX 0.6.4 on an Apple Silicon Studio with 256 GiB unified memory. Exact arithmetic, a forced weather-tool call with a Paris argument, and identification of a synthetic red image passed for this quant. Tool calls were checked for formatting, not executed. These are basic checks, not a full coding, vision or agent evaluation.
Greedy decoding reproduced severe repetition. With recommended sampling, six of seven requests finished naturally; one coding response hit the cap and contained a runtime error, and a story missed its requested length. This aggressive quant is experimental, not reliability-certified.
Five sampled coding requests delivered approximately 100–101 output tokens per second over API elapsed time. These are end-to-end observations, not controlled decode-only benchmarks; cache state was not controlled. Peak request memory and context-fit limits have not been measured, so no Mac memory-tier recommendation is claimed. Long-context, multi-turn and broader vision quality remain unverified.
Short test
With the sampling settings above, try:
What is 17 multiplied by 19? Answer with only the number.
The recorded arithmetic check returned 323 with temperature=0 and reasoning_effort="none".
A short correct response does not establish long-generation reliability.
Licence and attribution
The upstream repository declares Apache-2.0. Model training, architecture and the original Nex logo belong to Nex-AGI and the respective upstream contributors. This is an independent community conversion, not an official Nex-AGI release. Upstream benchmark scores are not evaluations of this quant.