Spark-X2.6-1.7B from SparkLLM/XHToken is an on-device chat model with about 1.7 billion parameters, free under Apache 2.0, with controllable thinking (none / low / high). On Ollama the q4_k_m tag is a 1.8 GB download. Our test on a CPU-only box with 8 cores and no GPU, using --think=low: it correctly said 91 is not prime (7 × 13) and named Canberra as Australia's capital, at about 25–27 tokens/s using about 1.9 GB of RAM; numbers measured on our test box, not typical performance.
Try it
ollama run SparkLLM/Spark-X2.6-1.7B:q4_k_m
Install Ollama from ollama.com first (v0.34.1 or later per the maker). The q4_k_m tag is a 1.8 GB download. Optional: add --think=low --verbose. We ran it on our CPU-only test box (8 cores, no GPU); speed and memory there are not typical performance.
Two independent, reliable sources for every claim. Every fact, number and claim in a story has to appear in at least two independent sources. Outlets that only repeat another outlet's report count as one source.
Quotes come from the primary source. We check every quote against the original statement, post, paper or document.
Unverified means removed. If something can't be confirmed, we cut it. We don't post it with a caveat.
Sources are linked. Every story lists the sources we used, so you can check them yourself.
Corrections
Spotted a mistake? Email corrections@chakkde.com with the story link and what you think is wrong. We review every report. If a story is wrong, we fix it and add a dated correction note to the story's page.