PAICE + NEAR Cascade Demo

AI Collaboration Assessment

Three-layer cascade — Chat, Middleware, Evaluation

Middleware layer

Cascade active: GPT OSS 120B (Chat) → Qwen3 30B (Middleware) → GLM 4.7 (Evaluation). All models run inside TEE enclaves via NEAR AI Cloud.

GPT OSS 120B · Chat Layer · Private

Welcome to PAICE! I'm here to assess how effectively you collaborate with AI. Let's explore how you approach problem-solving with AI tools. Could you tell me about a recent situation where you used AI to help with a work task?

Model Cascade

Chat Partner

GPT OSS 120B

Fallback: DeepSeek V3.1

Middleware (QA)

Qwen3 30B

Fallback: DeepSeek V3.1

Evaluation

GLM 4.7

Fallback: GPT OSS 120B

Integration Pipeline

✓ Connect to NEAR AI Cloud
2 Chat via TEE-protected cascade
3 Middleware QA check
4 Generate assessment scores (GLM 4.7)
5 Hash score payload (SHA-256)
6 Write attestation to NEAR testnet
7 Verify on-chain

Actions

Configuration

NEAR AI API Key

Contract Address

Network

Activity Log

[init] PAICE + NEAR Cascade Demo loaded

[cascade] Chat: GPT OSS 120B (Private)

[cascade] Middleware: Qwen3 30B (Private)

[cascade] Evaluation: GLM 4.7 (Private)

[config] Contract: paice-demo.testnet