AI Agent StoreXHToken · Download verified
Spark-X2.6 4B
Run a compact model locally for coding, text tasks and tool-assisted workflows. It builds on Spark X2.5 with additional training.
First reviewed 2026-10-21 · Checked 2026-11-03. The launch date has not been confirmed.
We still need comparable independent tests. You can compare the confirmed features and limits below, but we cannot yet give this version an Agent Fit or Practical Value rating. Try it on your hardware and tasks before relying on it.
What you can use today
You download the model files and supply the hardware and software to run them. Capacity is the amount of information a request can contain; settings and available memory can limit what fits.
Exact download ID: XHToken/Spark-X2.6-4B. Use the low or high thinking setting, or turn thinking off. Your application still needs to execute tools and manage permissions.
Available thinking settings: none, low, high. We do not yet have comparable independent results for these settings. The documented chat-template default is high when effort is omitted. A default is not a recommendation for every task.
Apache 2.0 weights. Use a compatible Spark runtime or plugin; local hardware and memory determine practical capacity. The repository existed October 19 and was updated October 20; an exact launch day is not established. The publisher now links a community browser demo using the INT4 weights. It was tested by its contributor on Windows with an NVIDIA GPU; that does not establish compatibility or speed on every browser or device.
The configuration and serving guide allow 1,048,576 combined tokens, subject to available memory. The example’s 131,072 output budget is not a verified universal output limit.
No hosted token price verified. Budget for your own hardware and operation.
Provider release and specifications ↗
Check availability and plans ↗
Test it before switching
Give each model the same examples, tools and definition of success. Record correct outcomes, response time, retries and total cost. Include the time and hardware needed to run it.
Compare caching and long-context costs ↗