Changelog

2026-08-26: New models in the picker

The recommended lineup is five models, with context up to 128k on the larger ones:

• Muse Glimmer 30B - reasoning
• Qwen3.5 9B - fast
• Qwen3.6 35B A3B - default chat
• Qwen3 Coder Next 80B - coding
• Qwen3.8 27B - agents. Qwen3.6 27B stays in the list if you want it.

GPT-OSS 20B and Qwen3.5 4B remain selectable. Model ids never remap. If you picked 3.6 last week, you still get 3.6.

Open them in the app, or as name:tier on the API.

2026-07-03: Redesigned app, public docs, and a thinking toggle

The web app is redesigned top to bottom: a cleaner sidebar, a calmer color palette, and a chat composer that stays out of your way. Same encryption, same workflow, less visual noise.

There’s a public docs site now at docs.sporeintel.com. It covers getting started, running a node, how credits and pricing work, the end-to-end encryption design, and a full API reference for the OpenAI-compatible endpoint.

Models that support reasoning now have a thinking toggle, and the API accepts OpenAI’s reasoning_effort so you can dial how much a model thinks per request.