Pricing

Build locally for free. Upgrade to delegate work to October and keep it moving in the cloud.

Free — Build locally

Your complete local workspace for coding with agents. Free, with no credit card required. Unlimited local canvases and projects, local agents, terminals, live previews, your own subscriptions and API keys, and on-device dictation on supported Macs.

Pro — Turn goals into results

Recommended. $20 per person per month with monthly billing, or $16.67 per person per month with yearly billing. Billed as $240 $200/year. Save $40 (2 months free). Tell October what you want done. Use Runs to plan, build and review with your agents, through chat or voice. Everything in Free, plus Universal Chat, universal voice control with online speech recognition, phone access to your running computer, and real-time collaboration.

Preview monthly allowances: $3 assistant usage, 120 online voice minutes and $5 October CLI credits.

Max — Keep work moving

$60 per person per month with monthly billing, or $50 per person per month with yearly billing. Billed as $720 $600/year. Save $120 (2 months free). Everything in Pro, with October Cloud to run agents in October-managed cloud workspaces. Keep agents working when your laptop is closed. Read the included usage and limits for assistant usage, online voice, October CLI credits and cloud compute.

Preview monthly allowances: $8 assistant usage, 300 online voice minutes, $10 October CLI credits and 20 pooled workspace-hours. 4 vCPU and 8 GiB RAM per workspace; 2 workspaces, both able to run concurrently.

Give October a goal.

October picks the right coding agents to plan, build and review, connects them, and passes work and feedback between them. You approve the plan and get the result—without juggling tools or relaying messages.

Three illustrative examples of Runs:

Review: “Check this PR for bugs.”

October: “I’ll review the code, then get a second opinion.”

  1. One agent reviews the code
  2. Another checks the findings
  3. They resolve the feedback
2 issues found — Open report

Fictional work and evidence to illustrate a Run’s result.

A retry can create the same order twice

In this example, src/checkout.ts creates a fresh request key on retry. That can create a second order if the first request succeeded before timing out.

Reuse the original request key.

The retry case has no test

The example checkout tests cover success and rejection, but miss a retry after a timeout.

Add a test confirming that a retry creates only one order.

The first audit went to an independent reviewer, who requested a retry trace. The author added it and the reviewer confirmed the revised findings.

The report covers the captured PR revision. A reviewed audit does not mean the PR is safe or merged. Publishing remains your decision.

Plan & build: “Add Google sign-in.”

October: “I’ll have Fable plan it, then hand the build to Grok.”

  1. Fable 5.1 plans in Claude Code
  2. Grok 4.6 builds from the plan
  3. Fable checks the finished work
Google sign-in built — Open report

Fictional work and evidence to illustrate a Run’s result.

A reviewed plan, handed to Grok

Fable 5.1 in Claude Code plans the sign-in button, callback and session handling. Grok 4.6 reviews the plan before Fable confirms it. October passes the agreed plan to Grok to implement.

Review the example implementation and sign-in flow.

The build checked against the plan

In this fictional result, Fable checks the completed flow and the tests for sign-in, cancellation and an expired session, and confirms that the agreed plan is covered.

Inspect the changes and test evidence before publishing.

October carries the agreed plan into the implementation, then returns the changes and test evidence to Fable for verification. You do not have to relay instructions between agents.

These are illustrative model choices, not a required pairing. October uses your available agents and accounts. This example does not deploy changes or configure a real Google account.

Three agents: “Build a team dashboard.”

October: “I’ll have one agent plan, one build, and another check the result.”

  1. Fable 5.1 writes the plan
  2. Grok 4.6 builds the dashboard
  3. GPT-5.6 Sol reviews in Codex
Dashboard built — Open report

Fictional work and evidence to illustrate a Run’s result.

Three agents, one agreed plan

Fable 5.1 in Claude Code proposes the dashboard layout, team filters and empty states. GPT-5.6 Sol in Codex reviews it, Fable confirms the plan, and Grok 4.6 builds it.

Explore the example dashboard and the changes behind it.

Feedback carried through to the finish

In this fictional review, Codex finds that the empty-team state is missing. October sends the correction to Grok, then brings the revised code back to Codex to check the fix and tests.

Review the final changes and test evidence before merging.

October keeps the plan, implementation and independent review together. It routes corrections to the right agent and brings the checked result back to the same Run.

These are example assignments. Agent, model and account choices can change with availability and your preferences. Nothing in this fictional Run has been published or merged.

How Runs work.

Separate accounts and allowances

October CLI credits cover October-funded model usage through the October CLI harness. They do not pay for separate agent subscriptions. Universal Chat, online voice and cloud compute have their own included allowances. Annual plans are paid upfront; included usage refreshes monthly. The same monthly allowances apply to annual plans. Included usage does not roll over and there are no automatic overage charges. Exhaustion affects only that service.

Download free or read the plan and offer details.