Pondero Brief: 2026-06-01: Copilot's token billing is live, and the per-model math matters

        June 1, 2026

Pondero Brief: 2026-06-01: Copilot's token billing is live, and the per-model math matters

Pondero Brief - JUNE 1ST, 2026

      Pro now includes $15 in credits, Pro+ jumps to $39/mo for $70 total. Here is the breakdown.

pondero.

                      BRIEF

                      ANALYSIS

GitHub Copilot switches to AI Credits billing today. Here is what it costs per model.

                  Pro's $15 credit pool vs. Cursor's flat $20/mo. We broke down the per-persona cost math for both tools.

                  JUNE 1ST, 2026 · BY JONATHAN HILDEBRANDT

GitHub Copilot switched to token-based AI Credits billing today. Every paid plan now maps to a dollar-denominated credit pool: Pro ($10/mo) includes $15 in total credits ($10 base + $5 flex allotment), Pro+ ($39/mo) includes $70, and a new Max tier ($100/mo) includes $200. One credit equals $0.01. Code completions stay unlimited on every paid plan. Model choice now drives your spend directly: Claude Opus costs $5/$25 per million tokens on Copilot, while GPT-4.1 costs $2/$8, per the GitHub billing docs.
Why it matters. Your model picker is now a budget dial. A heavy Opus session can burn through Pro's $15 allowance in a couple of hours. Cursor charges a flat $20/mo for 500 fast requests with no per-token metering, per the Cursor pricing page. We broke down the per-persona cost math for both tools.

See the per-persona cost breakdown →

          Anthropic closes $65B Series H at $965B post-money valuation.

        JUNE 1ST · PONDERO NEWSDESK

Announced May 28 alongside Claude Opus 4.8. At $965B post-money, per Anthropic's announcement, this is the second-highest-valued private tech company in the world. No IPO timeline was stated. The fundraise lands while Opus 4.8 pricing holds flat at $5/$25 per million tokens, same as Opus 4.7, per the Anthropic Opus product page.

          Gemini Spark rolls out to all US Google AI Ultra subscribers.

        JUNE 1ST · PONDERO NEWSDESK

Google's 24/7 ambient personal agent started reaching all Google AI Ultra subscribers in the US on May 29. Ultra costs $249.99/mo. Spark runs proactively in the background, checking your calendar, email, and search history and surfacing recommendations unprompted.

          GitHub Copilot Memory adds a repo-level off switch and CLI commands.

        JUNE 1ST · PONDERO NEWSDESK

The May 26 changelog update shipped a repository-level on/off toggle, /memory show and /memory off CLI commands, and clearer scope labels at capture time. With usage-based billing now live, stored repo context ties directly to premium model spend. Read our setup guide.

          Anthropic acquires Stainless, the typed SDK auto-generator.

        JUNE 1ST · PONDERO NEWSDESK

Stainless auto-generates typed SDK clients from OpenAPI specs and already built the Anthropic Python and TypeScript SDKs. The acquisition gives Anthropic direct ownership of the developer-experience layer for the Claude API as MCP client adoption grows.

          Make: automation platform that charges per operation, not per workflow.

        JUNE 1ST · PONDERO NEWSDESK

If you're building AI agent workflows and need a predictable bill, Make's per-operation model makes the math straightforward. Try Make.

          Cloudways: managed hosting for self-hosted AI tools from $15/mo.

        JUNE 1ST · PONDERO NEWSDESK

Managed cloud hosting for Open WebUI, n8n, and Flowise. Spin up a DigitalOcean server starting at $15/mo in about five minutes. Try Cloudways.

          Copilot or Cursor? The per-persona cost math after June 1 token billing.

        JUNE 1ST · PONDERO COMPARISON

Pro's $15 credit pool vs. Cursor's flat $20/mo, broken down for solo devs, team leads, and heavy agentic users. Read the Copilot vs. Cursor pricing breakdown.

          Opus 4.8 ships at the same $5/$25 per million tokens as 4.7. Upgrading is a one-line model ID swap.

        MAY 30TH · PONDERO REVIEW

Dynamic workflows, parallel subagents, and a faster output mode at the same price point. Read the Claude Opus 4.8 review.

          n8n bills per execution regardless of step count. That flips the decision at five-plus steps.

        MAY 27TH · PONDERO COMPARISON

Per-execution vs. per-step billing changes who wins at scale. We pulled the current pricing for all three. Read the n8n vs. Zapier vs. Make pricing guide. Try Make.

          Memory vs. copilot-instructions.md: what to let Copilot store and when to write the file yourself.

        JUNE 1ST · PONDERO GUIDE

The new repo-level off switch and CLI commands change the calculus. Here is when each approach fits. Read the Copilot Memory setup guide.

          Zapier ships two agent integrations: MCP for chat agents, TypeScript SDK for coding agents. Here is the line.

        MAY 31ST · PONDERO GUIDE

MCP gives your chat agent 7,000+ Zapier actions. The SDK gives your coding agent typed tool calls. Pick based on where the agent runs. Read the Zapier SDK vs. MCP guide.

Jonathan Hildebrandt

                      Co-founder and primary operator of Pondero. Writes the Pondero Brief.

Affiliate disclosure
                   · 
                  Unsubscribe
                   · 
                  Manage preferences

                  Pondero earns commissions on some links. This does not affect our editorial picks.

                                Don't miss what's next. Subscribe to Pondero AI:

                        What topics are you most interested in?

                        How did you hear about us?

            Email address (required)