Save your private key
Generate a key privately in the API portal and save it before registration. Email, personal name and Telegram are not required. Never put your secret into public configuration or a URL.
Prepare HESOYAM as your OpenCode provider.
Prepare your privately held key for the planned AInferen OpenAI-compatible endpoint; its new domain is not yet verified. This page provides configuration templates and model discovery; it is not a built-in OpenCode integration or a verified coding-agent deployment.
The current demo checks the connection
hs-mini uses Qwen2.5-0.5B q4_0: short text, limited quality, no tools or vision. It is not suitable for OpenCode agentic coding. Paid coding models and crypto top-ups remain closed until supplier and pricing verification.
Limits: 5 attempts/key/24h, 5 per registration network, 50 anonymous attempts globally; 6 messages, 1400 UTF-8 bytes and 128 output tokens. Failed generation consumes an attempt.
Runtime registration and limitsThree steps before connecting
Inspect the model and limits
Read /v1/registration and /v1/models first. Select the exact advertised identifier and model capabilities. The current hs-mini is not a coding model.
Prepare your version’s configuration
Use the matching V1 or V2 template and a private environment variable. Run a coding workflow only after a verified compatible model is available; these examples have not passed an OpenCode runtime test.
Configuration templates
These templates follow official documentation. They are configuration data; reading this page does not register accounts, make payments or call inference. The secret is read from a local environment variable; HESOYAM’s current limits still apply.
V1 uses provider / npm / options. V2 docs use providers / package / settings. The current published config.json schema is V1; the V2 template is docs-only and not runtime-tested. Do not mix the formats.
OpenCode may add system/project context and retry failed requests. This can exceed 6 messages/1400 UTF-8 bytes or consume another attempt. Configuration limit fields do not override API limits; unattended coding requires a separate compatibility test.
Crypto funding follows verified supply
The paid model, price and supplier must be verified before an invoice opens. HESOYAM’s proposed price ladder and own-brand router packages are described separately; no paid tariff is active today. A Launch setup may include one agreed OpenAI-compatible integration once its capabilities and acceptance are confirmed.
Router packages and proposed pricingDiscuss an OpenCode workflow
Share your OpenCode version, required model, tools, context and budget. Do not send API keys. Compatibility and scope are agreed first; a request does not create a payment.
Open support in Telegram