Help Center

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