Roadmap

TokenBroker is a living service, shaped as much by what testing reveals as by what was originally planned — see Tested & verified for examples of exactly that happening. Here’s the direction of travel, in rough order of priority:

Just shipped

• Mobile workers (Android & iOS) — contribute compute from a phone, not just a desktop.
• External cloud provider support (Google AI Studio) — a worker can now contribute a free-tier API key it holds personally as a second, independent capacity pool alongside Ollama Cloud.
• Live network status page at /network — real-time aggregate capacity, visible to everyone.

Now

• Expanding the models catalog toward nearly every open model — more Hugging Face imports (distills, specialists, select uncensored variants) and more community-served capacity.
• Re-adding very large models (certain GLM, Kimi, MiniMax and Mistral Large releases) as sustainable access becomes available.
• Improving worker onboarding and earnings transparency.
• More external cloud providers as verified free-tier keys become available.
• A real measured probe for Google AI Studio’s throughput, the same deliberately bounded method already used for Ollama Cloud — see Tested & verified for why the current figure is still labeled an estimate.

Next

• Bigger-model splitting across multiple GPUs for large jobs.
• More worker telemetry (uptime graphs, request-rate overlays, storage pool health).
• Refined credit value model tied to real network economics.
• Mobile-friendly dashboard polish.

Later

• A compute marketplace with explicit supply/demand pricing.
• Team/organization accounts with shared billing.
• More payment methods and automated invoicing.

The roadmap is a conversation, not a contract. If there’s a feature you need, tell us — the network is built by its members.