Default vs UNITIQX Extension

Public development specification generated from the live, bounded local inference worker. Baselines are representative UX expectations, not a claim that native commercial code, services, or user interfaces were copied.

The 12 model-backed experiences

Qwen

Actual model: qwen3:1.7b

Built: public preview, local text inference, short context replay, browser-local history.

Not built: streaming, files, voice, tools, provider identity, feature parity.

DeepSeek

Actual model: deepseek-r1:1.5b

Built: public preview, local text inference, short context replay, browser-local history.

Not built: streaming, files, voice, tools, provider identity, feature parity.

Copilot

Actual model: qwen2.5:1.5b

Built: public preview, local text inference, short context replay, browser-local history.

Not built: streaming, files, voice, tools, provider identity, feature parity.

Claude

Actual model: gemma3:4b

Built: public preview, local text inference, short context replay, browser-local history.

Not built: streaming, files, voice, tools, provider identity, feature parity.

Mistral

Actual model: granite3.3:2b

Built: public preview, local text inference, short context replay, browser-local history.

Not built: streaming, files, voice, tools, provider identity, feature parity.

Perplexity

Actual model: granite3.3:2b

Built: public preview, local text inference, short context replay, browser-local history.

Not built: streaming, files, voice, tools, provider identity, feature parity.

GLM

Actual model: qwen2.5:1.5b

Built: public preview, local text inference, short context replay, browser-local history.

Not built: streaming, files, voice, tools, provider identity, feature parity.

Gemini

Actual model: gemma3:4b

Built: public preview, local text inference, short context replay, browser-local history.

Not built: streaming, files, voice, tools, provider identity, feature parity.

Kimi

Actual model: qwen3:1.7b

Built: public preview, local text inference, short context replay, browser-local history.

Not built: streaming, files, voice, tools, provider identity, feature parity.

ChatGPT

Actual model: llama3.2:1b

Built: public preview, local text inference, short context replay, browser-local history.

Not built: streaming, files, voice, tools, provider identity, feature parity.

Grok

Actual model: qwen3:1.7b

Built: public preview, local text inference, short context replay, browser-local history.

Not built: streaming, files, voice, tools, provider identity, feature parity.

SPHYNX PRIME

Actual model: qwen2.5:1.5b

Built: public preview, local text inference, short context replay, browser-local history.

Not built: streaming, files, voice, tools, provider identity, feature parity.

Acceptance requirements

  1. Independent live reply from each mapped model
  2. Streaming delivery and cancellation
  3. Export/import and deletions
  4. Tested web research citations
  5. Scoped staging file edits and rollback
  6. Authenticated privileged production tools
  7. Accessibility and multiple-browser QA