Anthropic Boosts Claude Code Limits, Hands Out $100 Credits as Fable 5 Promotional Limbo Ends

Anthropic ended the Fable 5 promotional limbo on July 20 with a two-tier split: Max and Team Premium subscribers keep permanent access to the company's flagship model at 50% of standard weekly limits, while Pro and Team Standard users lose bundled access entirely and receive a one-time $100 usage credit. At the same time, the company extended its Claude Code weekly limit boost (+50% through August 19) and Claude Cowork 5-hour limit doubling (through August 5) — a clear signal that demand pressure and competitive dynamics are shaping Anthropic's product decisions in real time.

What Actually Changed on July 20

The Fable 5 situation has been a mess from the start. Anthropic originally scheduled the model's subscription exit for July 7, pushed it to July 12 after backlash, then again to July 19. Each extension was framed as a goodwill gesture, but the subtext was always capacity — Anthropic simply couldn't afford to keep Fable 5 open to everyone.

The final resolution is a compromise that satisfies nobody fully:

• Max ($100/mo) and Team Premium subscribers — permanent Fable 5 access, but capped at 50% of standard weekly usage limits. The cap arrives on the same day the 50% Claude Code weekly limit boost expires, creating a double squeeze on usable capacity.
• Pro ($20/mo) and Team Standard subscribers — lose bundled Fable 5 access entirely. Instead, they get a one-time $100 usage credit, expiring September 17.
• The $100 credit is less generous than it sounds. At Fable 5's API pricing ($10/M input, $50/M output tokens), a single long agentic session — say, a code migration or document analysis run producing 2 million output tokens — eats the entire credit. For Pro users who relied on Fable 5 for daily coding work, this is a one-shot consolation prize, not a sustainable solution.

The Promotions That Are Still Running

Anthropic quietly extended two existing promotions while the Fable 5 drama played out:

Claude Code weekly limits: +50% through August 19. Originally launched May 13, this promotion increases weekly usage limits for Pro, Max, and Team plans (and legacy seat-based Enterprise) by 50%. The 5-hour rolling window limits are not affected. The extension through mid-August gives developers breathing room — but the end date is already set.

Claude Cowork 5-hour limits: doubled through August 5. Launched June 5 and extended past the original July deadline, this one doubles the 5-hour usage window for Claude Cowork across the same eligible plans. Weekly limits are unchanged.

Both promotions apply automatically — no signup required. But both have hard expiration dates, and neither addresses the fundamental problem: demand for Anthropic's models is outstripping what the company can afford to offer at subscription prices.

Why This Is Happening: The Competitive Squeeze

The timing is not coincidental. July 2026 has been one of the most competitive stretches in the AI industry's history:

• Grok 4.5 (xAI, July 8) — Cursor-native, $2/$6 per million tokens, 112 tokens/sec output. An 80% cost reduction versus Fable 5 on per-task pricing.
• Kimi K3 (Moonshot, July 17) — 2.8-trillion-parameter open-weight model, $0.30/M cached input, full weights dropping July 27. The first 3T-class open model ever.
• GPT-5.6 Sol (OpenAI, July 9) — $5/$30 per million tokens, statistical tie with Fable 5 on the Artificial Analysis Intelligence Index (59.9 vs 58.9), at roughly one-third the per-task cost.
• GLM Coding Plan (Zhipu AI) — $3/month entry price, GLM-5.2 model with 1M context, free MCP tools. The cheapest path to a capable coding model.
• MiniMax M3 — $10/month entry, 1M context, 100+ tokens/second on high-speed plans.

The market is flooded with alternatives that are 80–95% as capable as Fable 5 at a fraction of the cost. Anthropic's pricing ladder now spans an order of magnitude: Haiku 4.5 at $1/$5, Sonnet 5 at $2/$10 (introductory through August 31), Opus 4.8 at $5/$25, and Fable 5 at $10/$50 per million tokens. The 2:1 output cost gap between Fable 5 and Opus 4.8 is the largest within Anthropic's own lineup.

Fable 5 is the best model on the market by most benchmarks — but it's also the most expensive to run, and the competition is closing the gap fast.

What the $100 Credit Actually Buys

Let's do the math. At $50 per million output tokens:

• A single code review session (200K output tokens) = $10
• A full code migration (1M output tokens) = $50
• A complex agentic task (2M output tokens) = $100 — the entire credit, gone in one session
• Light daily use (100K output tokens/day) = 20 days of coverage

The credit is a generous gesture, but in practice it's a bridge to nowhere. After it's exhausted, Pro users either pay out-of-pocket at API rates or switch to a cheaper model — either within Anthropic's lineup (Sonnet 5, Opus 4.8) or from a competitor.

Anthropic hasn't said when the credits will be distributed or what the expected usage guidelines are. The credit expires September 17, 2026.

The Bigger Picture

Anthropic is in a difficult position. Fable 5 — part of the new Mythos class — uses adaptive reasoning that demands substantially more compute per token than Opus 4.8. The 1-million-token context window and 128,000-token maximum output are designed for multi-hour autonomous sessions that burn tokens at scale. A mandatory 30-day data retention policy on all Mythos-class traffic adds operational overhead that Opus models don't carry.

The company can't afford to give Fable 5 away, but it can't afford to lose subscribers to competitors either. The $100 credit and extended limit promotions are stopgap measures — buying time while Anthropic scales infrastructure to meet demand.

A Claude Code lead engineer promised publicly in early July that the company aims to restore Fable 5 to subscriptions as soon as capacity allows. July 20 was not that date. Whether Fable 5 eventually returns as a standard Pro-tier benefit depends entirely on how fast Anthropic can build out compute.

For now, the message is clear: if you're on a Pro plan and you rely on Fable 5, start planning your migration. The $100 credit will buy you a few weeks at most, and after that, the API pricing is prohibitive for casual use.

---

For context on the current frontier model landscape, see our previous coverage: Grok 4.5: xAI's Cursor-Native Frontier Model Is Here and Kimi K3: Moonshot's 2.8T Open Frontier Model Drops.

Sources: Anthropic Support — Claude Code Weekly Limits Promotion · Anthropic Support — Claude Cowork Usage Promotion · Technobezz · Claude (@claudeai) on X — July 18, 2026 · CodingPlan.org — AI Coding Plan Comparison