Claude Cowork 2x Usage vs Fable 5 Promo (2026)

If you only read one section, read this one.

There are two promotions. They are not the same promotion.

Promo #1 — Cowork 2x usage. Doubled usage for Claude Cowork specifically. Window: June 5 through August 5, 2026, ending 11:59 PM Pacific. Applies automatically to Pro, Max, Team, and legacy seat-based Enterprise plans. Excludes Free plans and consumption-based Enterprise seats. Chat limits and overall weekly plan limits are untouched — and Claude Code is not part of this 2x, though it has its own quieter offer, covered below.

Promo #2 — Fable 5 promotional access. A separate lever: eligible users could spend up to 50% of their weekly subscription limits on the Fable 5 model at no extra cost, across Claude surfaces. That included window ended July 12, 2026 — after a last-minute extension from the original July 7 cutoff — and heavy Fable 5 use on paid plans now bills as usage credits at API rates. If you want the full anatomy of that cliff, my Fable 5 usage-limits breakdown covers it turn by turn.

And the third thing — the one I only discovered by reading my own usage panel while researching this post: Claude Code limits are 50% higher through August 19. It's right there in the boost notice in Settings → Usage, on a different percentage (50%, not 100%) and a different clock (August 19, not August 5) than the Cowork offer — and I've seen almost no coverage of it anywhere. Not doubled, but boosted; a third offer hiding in plain sight.

So when someone says 'Claude doubled limits': Cowork, yes, until August 5. Claude Code, half-boosted until August 19. Everything else, no. And the Fable 5 offer isn't 'still running' — it already changed shape. Every decision in the rest of this article follows from keeping those three facts separate.

The Cowork promo is simple once you scope it correctly — the entire problem is that almost nobody scoped it correctly.

What it is: doubled usage for Claude Cowork, Anthropic's agentic knowledge-work product — the one that operates on your files and folders, builds documents and spreadsheets, and coordinates sub-agents on multi-step tasks. The promo landed alongside Cowork's biggest platform expansion to date: on July 7, Anthropic announced Cowork is moving beyond the desktop app to the web at claude.ai and to mobile on iOS and Android, in beta, with Max subscribers first in line.

The scope, stated plainly: the doubled usage covers Cowork only. Chat limits and your overall weekly plan limits don't move. Claude Code is not part of the 2x either — but here the story picked up a wrinkle my own usage panel handed me mid-research: Claude Code carries its own separate boost, 50% higher weekly limits through August 19, stated right in the Settings → Usage notice alongside the Cowork line. Different percentage, different end date, near-zero coverage anywhere. So the accurate scoping is: Cowork doubled to August 5, Claude Code half-boosted to August 19, everything else standard. That's the detail the 'Claude doubled usage limits' headlines erased twice over — and it matters, because for most developers chat and Claude Code are where the bulk of weekly usage actually goes.

Eligibility: Pro, Max, Team, and legacy seat-based Enterprise plans, applied automatically — no opt-in, no code, no toggle. Two exclusions: Free plans, and consumption-based Enterprise seats (which bill by usage anyway, so a '2x included' framing wouldn't map onto them).

The window: June 5 through August 5, 2026, closing at 11:59 PM Pacific Time. Note the shape of that window — it opened a month before the web/mobile announcement made it famous. If you were using Cowork on desktop in June, you were already inside the promo without necessarily knowing it. On my own Pro plan, the boost notice sits at the top of Settings → Usage, and my test tasks during the window (details in the playbook section below) drew between 1% and 4% of a session each — noticeably gentler than the horror stories, though my project and task scoping did a lot of that work.

The second offer is the one this site has already covered in depth, so here's the compressed version with the July 20 status.

The Fable 5 promotional access let eligible subscribers spend up to 50% of their weekly subscription limits on Fable 5 — Anthropic's first Mythos-class frontier model — at no extra cost, across Claude surfaces. It was the mechanism that made a frontier model feel 'included' in ordinary plans for a few remarkable weeks.

The key event: that included window closed on July 12, 2026. Anthropic extended it from the original July 7 cutoff with hours to spare — a small drama my usage-limits post documented in real time — but the extension bought days, not months. Since July 12, heavy Fable 5 use on paid plans bills as usage credits at API rates. Fable 5 didn't disappear; it got a meter — and whether those credits are worth spending is its own question, one I answered with real code in Fable 5 vs Opus 4.8 on real CRM code.

Why this matters for the promo confusion: the Fable 5 offer and the Cowork offer overlapped in time during early July, both were 'Claude promotions,' and both had numbers attached (50%, 2x). When the July 7 Cowork news broke, the two blurred together in coverage — producing readers who believe the Fable 5 window runs to August 5 (it doesn't; it's already closed) and readers who believe their whole plan doubled (it didn't; only Cowork did).

My own post-cliff pattern, for context — and it's a confession as much as a data point: my usage-credit meter reads $10.47 spent (21% of my credit allocation) as of this week, and the bulk of it went to Fable 5 drafting work that, in hindsight, Sonnet 5 handles comfortably at zero credit cost. Roughly $2.50 of it was two blog drafts alone. Even the person who wrote the usage-limits explainer burned credits on the wrong tier — which is exactly why the strategic logic from my Sonnet 5 production-code testing bears repeating: the default tier absorbs the daily volume, and Fable 5 credits are specialist hours you spend deliberately, not a default you drift into.

Here's the table I wish every July 7 news story had included. Screenshot it, send it to whoever on your team is still confused, and move on with your week.

Since the July 7 announcement I've seen the same three wrong conclusions repeatedly — in comments, in team chats, and in at least one published guide. Clearing them up is half the value of this post.

Misreading #1: 'My Claude Code limits doubled.' No — but this one comes with a correction I owe you, because the simple version I almost published ('Claude Code is unchanged') is also wrong. Claude Code is not part of the Cowork 2x. It has its own separate boost: 50% higher weekly limits through August 19, stated plainly in the Settings → Usage notice on my Pro plan and covered almost nowhere. So if you're running long agentic coding sessions expecting doubled headroom, you have half of that — 50%, not 100% — and on a clock that runs two weeks longer than the Cowork one. Plan sprints against the real numbers: half-boosted until August 19, standard after.

Misreading #2: 'My weekly plan limits doubled.' Also no. The weekly limits that govern your overall plan are explicitly untouched. What the 2x does is make Cowork usage draw down your allocation at effectively half the rate during the window. That's a real and valuable thing — Cowork tasks are token-heavy, with multi-step reasoning, file operations, and sub-agent coordination consuming the equivalent of many regular chats per task — but it's a different thing from a bigger plan.

Misreading #3: 'The Fable 5 promo runs to August 5 too.' This is the merge error in its purest form. The August 5 date belongs to the Cowork promo. The Fable 5 included window ended July 12. If you've been casually using Fable 5 this week on the assumption it's still inside a promo window, your usage panel has news for you.

One honest observation about why this confusion was so easy to fall into: both promos were genuine, both were generous, and both were announced in a compressed news window where 'Claude usage' was the shared headline noun. This isn't a story about misleading messaging — the source documentation scopes everything precisely. It's a story about what happens between precise documentation and fast coverage, which is exactly the gap this site exists to fill.

Untangling the promos is table stakes. Here's the part that pays for the read: the 2x window is a free measurement instrument, and it expires in about two weeks.

The core problem it solves: nobody knows what Cowork actually costs on their plan until they've run real work through it. Cowork tasks are expensive by design — a single task can involve multi-step reasoning, dozens of file operations, and parallel sub-agents, and users consistently report that heavy daily Cowork use draws down Pro allocations fast. Auto mode adds further overhead: it runs parallel safety classifiers, which consume more tokens than approving steps manually. In my own testing on a client project (a 3D real-estate CRM), the spread between tasks was the real lesson: a one-page Word property summary cost +1% of a session, a supervised folder-cleanup plan +2% — but the Excel workbook build, which had to trawl the project's data files and assemble multiple sheets, cost +4%, four times the document task. Same product, same model, same project; the task shape drove the cost.

So here's the playbook I'm running myself.

One: pick your realistic weekly Cowork workload — not a demo, the actual mix. For me that's client document generation, spreadsheet builds from ERP exports, and the folder-cleanup tasks I've been meaning to automate. Two: run it inside the 2x window and record what your usage panel shows. Effective cost during the promo is half of normal, so double what you observe to get your post-August-5 number. My three test tasks drew a combined ~7% of session usage and moved the weekly bar one point (41% to 42%) during the promo — doubling that for post-promo math, a realistic week of eight to ten similar tasks looks entirely sustainable on Pro for my workload, with the caveat that my tasks were scoped, supervised, and run on a modest-sized project; heavier Auto-mode use on a large codebase is a different animal. Three: decide with that number, not vibes. If doubled-observed-usage fits comfortably in your plan, Cowork graduates to a permanent slot in your week. If it doesn't, you now know before the promo ends — and you can either trim the task list, batch work into fewer larger tasks, or step through approvals manually instead of Auto mode to cut the classifier overhead.

Four — and this is the one people skip: check whether the tasks you're feeding Cowork actually need Cowork. Some of my early Cowork tasks were really just chats with a file attached, and chats don't burn Cowork usage at all. The promo window is also the cheapest time to learn where that boundary is for your work.

On August 6, nothing dramatic happens — and that's worth saying plainly, because 'promo ends' headlines will make it sound like a cliff. Cowork keeps working on every eligible plan. Your usage simply counts at the normal rate again. There's nothing to cancel and no decision forced on you.

The more interesting thing is the pattern this is part of. In the span of six weeks, Anthropic shipped a near-Opus default model (Sonnet 5, June 30), closed the Fable 5 included window (July 12), and ran a doubled-usage promo on Cowork while expanding it to web and mobile (July 7 – August 5). Read together, that's a coherent strategy: make the default tier genuinely capable, meter the frontier tier, and use time-boxed promos to teach users what the agentic products are worth before charging full freight for them.

For users, the practical lesson is the same one I drew in the Sonnet 5 post, now with a second data point: model and product access is a planning variable, not a constant. Included windows close. Promos expire. Extensions happen with hours to spare. The stable strategy isn't chasing each offer — it's measuring your own usage during the cheap windows so that every pricing change lands on a spreadsheet instead of a surprise.

That's what the next two weeks are for. The promo untangling you needed is above; the measurement habit is what you should actually take away.