Claude Max and Team API Credits (2026): $100–$500 a Month and How to Claim

Claude Max and Team plans now include monthly credits for the Claude API: $100 on Max 5x, $200 on Max 20x and up to $500 pooled on Team. You have to claim them once, they cover your own apps and agents rather than Claude Code, and they expire every billing cycle. All plan details and prices below are as of October 17, 2026, taken from Anthropic’s help center, pricing pages and Claude Code changelog.

What are Claude’s monthly API credits?

Claude’s monthly API credits are Claude Platform credits included with Max and Team subscriptions, for building and running your own apps and agents on the Claude API. They sit apart from your chat and Claude Code limits.

Anthropic announced the credits on October 7, 2026, together with Claude Haiku 5.5, and rolled them out to subscribers over the following days. Anthropic’s pricing page now lists “$100 to $200/month in API credits” under Max and up to $500 a month in pooled API credits under Team. Free, Pro and Enterprise plans don’t get them.

The credits are an addition, not a trade. Your usage limits in Claude, Claude Code and Cowork stay exactly as they were, the help center says. They also aren’t the Agent SDK credit Anthropic announced in June: that offer isn’t available, according to the same article.

How much API credit does each Claude plan get?

Max 5x gets $100 a month, Max 20x gets $200, and Team plans get $20 per Standard seat and $100 per Premium seat, pooled and capped at $500.

How the Team pool works, according to Anthropic’s help center:

• One balance per team. Three Standard seats and two Premium seats give $260 a month. Adding a third Premium seat raises the next cycle to $360.
• The cap is $500. Larger teams still get $500, the most any Team plan receives.
• Discounted Team plans count too. Nonprofit and Scientists Team plans get the same $20 and $100 per seat.

If you move from Max 5x to Max 20x, you receive prorated credit right away and $200 each cycle after that. Our Claude Pro vs Max comparison covers what else changes between the tiers.

How do you claim Claude API credits?

You claim Claude’s API credits once by linking a Claude Console organization to your plan in claude.ai’s billing settings. After that, credits arrive in that organization every billing cycle.

1. Wait until you have been on Max or Team for seven days. New subscribers can’t claim earlier.
2. On claude.ai in a web browser, open Settings > Billing (Max) or Organization settings > Billing (Team).
3. In the API credits section, select Link organization, then choose or create the Console organization that should receive the credit.
4. Review the Supplemental Credit Terms and confirm.
5. Open the Console’s billing page, find the credit under Promotional credits, and create an API key.

A few rules to know before you click:

• Roles. On Max you must be the subscriber; on Team, an Owner or Primary Owner. In the Console organization you need the Owner, Admin or Billing role.
• One link, fixed. Each plan links to one Console organization, and each organization takes credits from one plan. You can’t change the link yourself, so pick the organization you will build in.
• No card needed. You don’t have to add a payment method on the Claude Platform to claim or use the credit.
• Mobile subscribers. Plans bought on iOS or Android qualify, but the claim happens on claude.ai in a browser.
• Same email helps. Anthropic suggests signing in to the Console with the same email you use on claude.ai.

What do the API credits cover, and what don’t they?

Claude’s monthly API credits cover the Claude API, Claude Managed Agents, the Agent SDK and the Console Playground. They don’t cover interactive Claude Code, extra usage in the Claude apps, or Claude on other clouds.

Covered, with any available Claude model:

• The Claude API, both the Messages API and the Message Batches API
• The Playground in the Claude Console
• Claude Managed Agents, Anthropic’s hosted agent harness, now in beta
• The Claude Agent SDK, and claude -p when you run it with an API key from the linked organization

Not covered:

• Interactive Claude Code in the terminal, an IDE, the desktop app or the web
• Extra usage (usage credits) in Claude, Claude Code or Cowork
• Claude on Amazon Bedrock, Google Cloud Vertex AI, Microsoft Foundry or Claude Platform on AWS

How the balance behaves: credits refresh each billing cycle and monthly on annual plans, expire at the end of the cycle with no rollover, and are spent before any credits you bought. Everyone with an API key in the linked organization draws from the same balance, so Team owners can set workspace spend limits in the Console. When the credits run out and the organization has no purchased credits, API requests stop until the next deposit; nothing is charged to your Claude plan.

What can $100 of Claude API credit buy?

At Anthropic’s list prices, $100 buys about 1 billion Haiku 5.5 input tokens on short prompts, 50 million Sonnet 5.5 input tokens, or roughly 170 one-hour Managed Agents sessions on Opus 5.5.

Our arithmetic, from Anthropic’s API pricing page:

• Haiku 5.5: $100 covers 1 billion input tokens or 200 million output tokens on prompts up to 100,000 tokens.
• Sonnet 5.5: $100 covers 50 million input tokens or 10 million output tokens.
• Managed Agents: sessions bill tokens at model rates plus $0.08 per session-hour while running. Anthropic’s own example, a one-hour Opus 5.5 session with 50,000 input and 15,000 output tokens, costs $0.58, so $100 pays for about 172 of them.

Real bills depend on prompt length, caching, effort level and tools. Web search inside a session costs $10 per 1,000 searches, and fast mode doesn’t work with the Batch API.

What changed in Claude’s API prices in October 2026?

Three price changes landed in October 2026: Claude Haiku 5.5 launched on October 7, Sonnet 5.5 cache reads fell by half the same day, and fast mode for Opus 5.5 became generally available.

• Claude Haiku 5.5. Anthropic calls it the cheapest, fastest and most capable small model it has released. It costs 90% less than Haiku 4.5 on prompts up to 100,000 tokens and 50% less above that, which Anthropic says works out to about 75% cheaper on average. It has a 1M-token context window, up to 128K output tokens and, for the first time on a Haiku model, an adjustable effort setting. Anthropic pitches it for summaries, classification, compaction and subagent work, according to its Haiku 5.5 announcement.
• Sonnet 5.5 cache reads. They dropped from $0.20 to $0.10 per 1M tokens on October 7. Because cache reads make up a large share of agent traffic, Anthropic estimates Sonnet 5.5 now runs about 20% cheaper on most agentic work.
• Opus 5.5 fast mode. Setting "speed": "fast" on a request gives up to 2.5x higher output tokens per second at twice the standard price, $8 input and $40 output per 1M tokens. Opus 5.5 no longer needs a beta header or an access request, according to the fast mode docs. It runs on the first-party API and on Microsoft Foundry.

In Claude Code, you can switch to Haiku 5.5 with /model haiku. That changes how fast you use your plan’s limits, not your API credits.

Which Claude Code updates help your usage limits last longer?

Four Claude Code changes from October 2026 save usage: idle compaction, per-subagent effort and model choice, new /usage warnings and a web search budget that refills. All of them work within your existing plan limits.

• Idle compaction (version 2.1.299, October 14). A terminal conversation over 200K tokens that sits idle for about 54 minutes is compacted just before its one-hour prompt cache expires. When you come back, Claude reprocesses the summary instead of the whole conversation. Set "idleCompaction": false in settings.json, or use the Idle compaction row in /config from version 2.1.302, to turn it off.
• Effort and model per subagent (2.1.292, October 6). You can ask in your prompt for low-effort subagents to search a codebase and one high-effort subagent to judge the results, or for Haiku on quick lookups. Before, that needed a custom subagent definition.
• Usage warnings (2.1.299). /usage now flags max effort, workflows, always-on context and large tool results, the usual causes of a fast-draining session.
• Web search budget (2.1.290, October 5). Interactive sessions now get 100 searches an hour that refill, instead of stopping after 200.

Anthropic’s engineers add one prompt tip: subagents optimize for thoroughness by default, so one sentence in your prompt saying that time matters makes them pace themselves faster, with what Anthropic calls a reasonable balance between speed and quality. For how 5-hour sessions and weekly caps work, see our guide to Claude usage limits, and check Claude Watch before long chat sessions on weekdays.

What’s new in the Claude Code desktop app and cloud sessions?

The Claude Code desktop app got a new session screen, saved environments and nested sessions in October 2026, and cloud sessions now pick up your local work exactly as you left it.

Desktop app, according to Anthropic’s Claude Code newsletter and desktop docs:

• New session screen. Recent folders are one click away, and you can save a setup as an environment, such as a few folders, an SSH host or the cloud.
• Nested sessions. Ask Claude to run tasks in parallel and each gets its own session that reports back to your main one when done. Suggested follow-up work can start in its own session with one click.
• Side questions in a tab. /btw now opens a side question in its own tab.
• Status at a glance. The sidebar shows which sessions are still working and which need you, and a side rail shows where a session runs, its context and its changes.

Cloud sessions, according to Anthropic’s cloud docs:

• Continue locally started work. Moving a local session to the cloud now carries your uncommitted changes and unpushed commits, so you don’t have to commit or push first. On macOS, Linux and WSL, uncommitted changes to files that look like credentials, such as .env files and private keys, stay on your machine.
• Network secrets on Team. Team plans can now add an API key to a cloud environment, as Pro and Max already could, and Claude calls the listed hosts without seeing the key.

If you claimed the Claude Code cloud credit before October 7, spend it on cloud sessions before it expires on November 4, 2026. Our AI coding tools comparison shows how Claude Code’s pricing compares with Cursor, Codex and Copilot.

Are there free Claude offers for open-source maintainers?

Yes. Claude for Open Source gives approved maintainers 6 months of Claude Max 20x for free, and since October 8, 2026, Anthropic’s OSS Scanner runs free security scans for established open-source projects.

Claude for Open Source is capped at 10,000 approved recipients, and you apply with your GitHub account. Applicants need to meet one of these bars, according to Anthropic’s program page and terms:

• Projects with 500+ dependent repositories, 100+ dependent packages or 200,000+ monthly downloads
• A listed committer or maintainer role on a recognized foundation or language project
• 100+ merged pull requests to repositories you don’t own in the last 12 months
• A repository with 20+ unique external contributors, or an OpenSSF criticality score of 0.4 or higher

The GitHub account must be at least two years old, with public open-source activity in the last 90 days. After 6 months, members who had a paid plan go back to it at their old rate unless they cancel; everyone else returns to the Free plan.

OSS Scanner is opt-in: core maintainers enroll by opening a pull request with a project config and an offline Dockerfile in Anthropic’s oss-scanner repository on GitHub. Projects then get periodic scans by Anthropic’s strongest models at no cost, with reports that haven’t been reviewed by a human. Acceptance follows criteria similar to Google’s OSS-Fuzz, so brand-new projects and most forks don’t qualify. OpenAI runs a similar program, Codex for Open Source, with six months of ChatGPT Pro.

Bottom line

Claim the credits if you are on Max or Team and might build anything on the API. Claiming costs nothing, but only support can change the organization link later.

• Max subscribers who build on the API: link the Console organization you actually use, since you can’t switch it yourself. $100 or $200 a month covers a lot of Haiku 5.5 and Sonnet 5.5 work.
• Max subscribers who only use Claude Code: the credits won’t extend your limits. Idle compaction, Haiku 5.5 subagents and the new /usage warnings will.
• Pro users weighing Max 5x: Max 5x costs $80 a month more than Pro and includes $100 of API credit, so it pays off fastest if you already spend that much on the API.
• Team owners: one owner claims for the whole team; set workspace spend limits so one project can’t drain the pool.
• Open-source maintainers: apply to Claude for Open Source and, for critical projects, OSS Scanner.

Our API credits deal page tracks any change to the terms, and every other Claude offer is on our Claude page and the promotions calendar.