Pulse Alpha Model: OpenCode's Free Stealth Release

What Pulse Alpha is

Pulse Alpha is a code name, not a product name. It is served by OpenCode as an OpenAI-compatible chat model that is free to call for a limited period while the operator collects feedback. That pattern is how several recent frontier models first appeared: anonymous listing, benchmark screenshots in developer communities, then an official name a few weeks later. Until that announcement, every capability claim about the model is provisional.

What is confirmed so far (2026-11-03)

This is the short list of facts that hold today. Anything not on it — parameter count, context window, benchmark scores, the lab behind it — is not public.

• Listed on OpenCode as Pulse Alpha Free, model ID pulse-alpha-free.
• Served through an OpenAI-compatible chat completions endpoint, so existing OpenAI SDK code works with a base URL and model ID change.
• Free of charge during the preview period, described by the operator as a stealth model used to gather feedback.
• The developer has not been disclosed. Forum guesses about which lab is behind it are unconfirmed.
• No public paid tier, no public SLA, no public data-processing commitment specific to this code name.

How to call Pulse Alpha (model ID pulse-alpha-free)

Get access through OpenCode first, then send a standard chat completions request. Keep the key on your server, and re-read the provider's documentation before production use: a preview model can be renamed, re-limited, or withdrawn at any time.

curl YOUR_PROVIDER_BASE_URL/chat/completions \
  -H "Authorization: Bearer YOUR_OPENCODE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"pulse-alpha-free","messages":[{"role":"user","content":"YOUR_TEST_PROMPT"}]}'

Need a paid, documented API instead?

A free stealth model is fine for evaluating output quality. It is a poor foundation for production: the name changes when the model is unveiled, limits move, and there is no commercial agreement behind it. If you are shipping, route the same request to a model with published rates and a stable identifier.

How Pulse Alpha compares to earlier stealth releases

Every code name in this table followed the same path: anonymous listing, free or discounted preview, then either an official name or a quiet shutdown. Treat the last row as the most likely outcome for any preview.

Dates and status are recorded when each page is updated. Open the linked page for current rates and the disclosed developer.

What to watch next

Three signals tell you the preview is turning into a real product, and one of them usually shows up first in developer communities rather than in an announcement.

• An official name appears. The listing switches from a code name to a vendor plus version number.
• A paid tier is published. Per-token rates and a rate-limit table mean the model is becoming a commercial offer.
• The free listing disappears. Silence is also an answer: some previews end without ever getting a name.
• Benchmarks get claimed by a vendor. Until the lab speaks, treat any score screenshot as unverified.

When the official name is published, this page will be updated to say which model Pulse Alpha turned out to be, and where it is available through TokenRa.

Before you send anything: data handling

Free previews are usually free because the operator wants evaluation data. With a code-named model there is no model card to check and no data-processing commitment published under that name. Treat every prompt as readable by the operator: do not send customer records, credentials, regulated data, or anything you would not put in a public issue. If your use case requires a data policy, use a model whose provider publishes one.

Pulse Alpha FAQ

Continue exploring