QMesh Support

Help Center

Find setup guides, operating boundaries, troubleshooting steps, and the right support channel

02

Guides

WEB · API · ERP · AGENT

How do I publish a task?

Choose a professional template, add inputs, security, budget, and acceptance rules, review the quote, then submit. Compute is allocated only after an execution plan is accepted

Open system demo
WINDOWS · LINUX · AGENT

What can a personal computer contribute?

CPU or GPU work such as inference, evaluation, simulation, compilation, parameter search, reconstruction, and verification. Physical quantum hardware remains with QPU providers

Open system demo
MATCH · REWARD · DEADLINE

Why do I not see matched tasks?

Complete Agent pairing, hardware checks, benchmark, security policy, availability, and minimum reward. Marketplace tasks appear only after compute is verified and eligible

Open system demo
FIT · TRUST · COST

How does QMesh choose a route?

QMesh compares workload fit, model and hardware capability, security zone, location, availability, deadline, expected quality, verification cost, and total accepted-result cost

Product
PROOF · POLICY · ACCEPT

How are results verified and settled?

QMesh binds input hashes, route, runtime, output, and verifier decisions into proof. Settlement starts only after policy checks or an authorized review accepts the result

Security
CACHE · BATCH · LOCAL

How can QMesh reduce API cost?

It uses caching, batching, rules, local or open models, and confidence gates for suitable work, and calls commercial APIs only when they add value. A pure API proxy does not reduce cost

Economics
COMMUNITY · TRUSTED · PRIVATE

Will sensitive code run on public computers?

Not by default. Sensitive work stays in trusted or private compute. Community compute receives only public or sanitized, bounded, verifiable work units

Security reports
RESPONSES · CHAT · EMBEDDINGS

Which models, Agents, and files are supported?

The current system supports OpenAI-compatible Responses, Chat Completions, and Embeddings; Windows and Linux Agents for Ollama, LM Studio, and vLLM; and task packages such as QASM, Python, QUBO, notebooks, ZIP or configuration, and structured data subject to policy

Open system demo
03

All questions

QMesh is a task-to-settlement control plane that connects task publishers, model APIs, QPUs, GPUs, HPC, edge, and contributed compute, then verifies accepted results

Choose a professional template, add inputs, security, budget, and acceptance rules, review the quote, then submit. Compute is allocated only after an execution plan is accepted

CPU or GPU work such as inference, evaluation, simulation, compilation, parameter search, reconstruction, and verification. Physical quantum hardware remains with QPU providers

Complete Agent pairing, hardware checks, benchmark, security policy, availability, and minimum reward. Marketplace tasks appear only after compute is verified and eligible

QMesh compares workload fit, model and hardware capability, security zone, location, availability, deadline, expected quality, verification cost, and total accepted-result cost

QMesh binds input hashes, route, runtime, output, and verifier decisions into proof. Settlement starts only after policy checks or an authorized review accepts the result

It uses caching, batching, rules, local or open models, and confidence gates for suitable work, and calls commercial APIs only when they add value. A pure API proxy does not reduce cost

Not by default. Sensitive work stays in trusted or private compute. Community compute receives only public or sanitized, bounded, verifiable work units

The current system supports OpenAI-compatible Responses, Chat Completions, and Embeddings; Windows and Linux Agents for Ollama, LM Studio, and vLLM; and task packages such as QASM, Python, QUBO, notebooks, ZIP or configuration, and structured data subject to policy

No. The public demo uses synthetic sample tasks and sample contracts, with no real customer execution or payment unless an authorized production environment is explicitly enabled

04

Contact and escalation

Send the right context to the right team. Never include passwords, API keys, private task data, or personal information

Before contacting support

  1. Record the task, compute, proof, or settlement ID
  2. Include your role, browser, operating system, and local time
  3. Attach a screenshot and the exact error text
  4. Remove credentials, API keys, personal data, and confidential inputs