Research the official provider defaults, inspect the distinct UNITIQX changes, and see exactly which endpoints and model tests exist. These are original provider-inspired applications; they are not logins to twelve commercial AI companies.
Research default: model chat and reasoning; Qwen interface has additional official platform features
UNITIQX specialization: architecture critique, schemas, deterministic acceptance plans
Actual inferred model: qwen3:1.7b
Last local backend test: passed · prior public proxy: passed
Missing: official provider inference, provider-native tools, browser streaming, true vendor feature/visual parity and unreviewed filesystem access.
Research default: reasoning and explanation in the official DeepSeek ecosystem
UNITIQX specialization: counterexamples, failure modes, falsifiable test instructions
Actual inferred model: deepseek-r1:1.5b
Last local backend test: passed · prior public proxy: passed
Missing: official provider inference, provider-native tools, browser streaming, true vendor feature/visual parity and unreviewed filesystem access.
Research default: IDE integration, repository-aware suggestions, agent editing subject to user permissions
UNITIQX specialization: original code review plans, integration test breakdowns
Actual inferred model: qwen2.5:1.5b
Last local backend test: passed · prior public proxy: passed
Missing: official provider inference, provider-native tools, browser streaming, true vendor feature/visual parity and unreviewed filesystem access.
Research default: conversation and side-by-side editable artifacts under Anthropic features
UNITIQX specialization: clear original documents and local drafts
Actual inferred model: gemma3:4b
Last local backend test: passed · prior public proxy: passed
Missing: official provider inference, provider-native tools, browser streaming, true vendor feature/visual parity and unreviewed filesystem access.
Research default: assistant chat and Mistral model tools as provided by the official platform
UNITIQX specialization: small-model efficiency checklists and scoped replies
Actual inferred model: granite3.3:2b
Last local backend test: passed · prior public proxy: failed_or_interrupted
Missing: official provider inference, provider-native tools, browser streaming, true vendor feature/visual parity and unreviewed filesystem access.
Research default: web-backed answer and citation workflows from the official platform
UNITIQX specialization: source verification plans; no invented live citations
Actual inferred model: granite3.3:2b
Last local backend test: passed · prior public proxy: failed_or_interrupted
Missing: official provider inference, provider-native tools, browser streaming, true vendor feature/visual parity and unreviewed filesystem access.
Research default: assistant interaction and developer model APIs in the official ecosystem
UNITIQX specialization: data schema critique, validation rules, compatibility planning
Actual inferred model: qwen2.5:1.5b
Last local backend test: passed · prior public proxy: failed_or_interrupted
Missing: official provider inference, provider-native tools, browser streaming, true vendor feature/visual parity and unreviewed filesystem access.
Research default: Google assistant app, content creation and multimodal features under official availability
UNITIQX specialization: concept exploration, UX critique and alternatives
Actual inferred model: gemma3:4b
Last local backend test: passed · prior public proxy: failed_or_interrupted
Missing: official provider inference, provider-native tools, browser streaming, true vendor feature/visual parity and unreviewed filesystem access.
Research default: official chat, context and document work depending on app/account
UNITIQX specialization: summaries and state tracking within bounded chat history
Actual inferred model: qwen3:1.7b
Last local backend test: passed · prior public proxy: failed_or_interrupted
Missing: official provider inference, provider-native tools, browser streaming, true vendor feature/visual parity and unreviewed filesystem access.
Research default: OpenAI ChatGPT platform supports projects, tools and other capabilities depending on plan/settings
UNITIQX specialization: general local chat, task analysis and browser-local saved sessions
Actual inferred model: llama3.2:1b
Last local backend test: passed · prior public proxy: failed_or_interrupted
Missing: official provider inference, provider-native tools, browser streaming, true vendor feature/visual parity and unreviewed filesystem access.
Research default: official Grok assistant and xAI model features depending on tier and product
UNITIQX specialization: skeptical design challenge and test-first critique
Actual inferred model: qwen3:1.7b
Last local backend test: passed · prior public proxy: failed_or_interrupted
Missing: official provider inference, provider-native tools, browser streaming, true vendor feature/visual parity and unreviewed filesystem access.
Research default: no third-party vendor default; built from original UNITIQX mission-control requirements
UNITIQX specialization: original architecture coordination, evidence and release criteria
Actual inferred model: qwen2.5:1.5b
Last local backend test: passed · prior public proxy: failed_or_interrupted
Missing: official provider inference, provider-native tools, browser streaming, true vendor feature/visual parity and unreviewed filesystem access.
| Capability | State | Implementation / boundary |
|---|---|---|
| Visitor can open a named public HTTPS chat workspace | implemented | Approved URL per identity; independently themed original UI |
| Original role-specific instruction prompt | implemented | Dedicated persona instruction injected by public local broker |
| Genuine bounded local CPU Ollama text response | implemented_if_quota_and_model_available | Model and fallback identified per reply, shared CPU concurrency 1 |
| Multi-turn with a bounded context replay | implemented | Client passes recent 6 messages up to 750 chars each |
| Browser-local conversations and JSON import/export | implemented | No public chat history database |
| Editable per-workspace Workpad and approved public-document read | implemented_pending_browser_release_qa | Local draft writes; only vetted static GET public sources |
| Anonymous filesystem or production write access | intentionally_disabled | Owner authorization required; not available in public chat |
| Commercial provider model and account sign-in | not_connected | No commercial vendor API billing or native tool parity |
| SSE token streaming and canceling inference | not_implemented | Current public API returns complete JSON on success |
| Live cited web research and arbitrary connector execution | not_implemented | No web tool behind this bounded inference service |
| Vision, audio, file upload to models, video generation | not_implemented | Not present in this public broker |
| Official feature/visual equivalence and human certification | unverified | Current UIs are original design variants, not identical vendor products |
A public project may be visible through a sanitized status record while its secret data, unreviewed code, authorization controls and draft documents stay restricted. That separation is intentional and preserves visitors and the server.
Read the complete machine-readable capability contract ↗ · Latest actual local-provider test evidence ↗ · Public project stages ↗