AI Releases & Pricing

July 18, 2026

Fable 5 Becomes Permanent on Max and Team Premium July 20, Pro Gets a One-Time $100 Credit

Subscribe
Get every issue
EmailIn your inboxPodcastAny playerRSSAny reader
Listen

Anthropic broke the three-week deadline pattern today: no fourth extension, no credits-only cliff. From July 20, Max and Team Premium keep Fable 5 at 50% of limits, Pro and Team Standard move to usage credits with a $100 one-time credit, and the Claude Code weekly boost ends. Gemini 3.5 Pro "launched July 17" posts are content-mill false reports, and SpaceXAI open-sourced Grok Build.

Anthropic ends the Fable 5 cliff drama with a permanent tier split

The deadline this feed has tracked through three wire-moves (Jun 23 to Jul 7 to Jul 12 to Jul 19) is done. At roughly 7:44 AM PT today, Claude's official account posted on X: "Beginning July 20, Claude Fable 5 will be included in all Max and Team Premium plans, at 50% of limits. Pro and Team Standard users will continue to have access to Fable via usage credits, and will receive a one-time $100 credit. Demand for Fable has been challenging to predict, which is why we rolled it out to subscription plans in stages, extending access several times as we secured additional capacity."

That is the new permanent structure, not another one-week reprieve. Max and Team Premium keep Fable 5 inside their subscription at 50% of weekly limits, indefinitely. Pro and Team Standard lose included access: after a one-time $100 usage credit, every Fable 5 token runs on prepaid usage credits at the API rate of $10 per million input and $50 per million output, double Claude Opus 4.8's $5/$25 and tied with Mythos 5 at the top of Anthropic's rate card.

A follow-up post acknowledged the whiplash: "We know this has been frustrating, and we want to give you more certainty about what your plan includes. We are making access standard at 50% usage for the plans that use Fable most intensively," per CNBC TV18 and The Economic Times. Anthropic said it is "continuing to invest in new capacity" and will "update Fable 5 access policies regularly based on available capacity," leaving the door open to changes for lower tiers later.

Fable 5 access from July 20
Anthropic's permanent tier split, announced July 18 2026. Source: @claudeai; platform.claude.com pricing.

Two fine-print items matter for anyone budgeting Fable this week. First, the separate Claude Code 50% weekly limit boost ends July 19 at 11:59:59 PM PT, so from July 20 Code weekly ceilings revert to standard, a roughly one-third cut from the promo ceiling, and Fable's 50% cap then applies to that lower base. Second, the $100 credit for Pro and Team Standard is a one-time grant, not a recurring allowance. At $50 per million output tokens, The Decoder notes it covers roughly two million output tokens and then expires. Free, standard Enterprise, usage-based Enterprise, and API users were already excluded from the promo and are unaffected. The Fable 5 API price stays $10/$50, with prompt-caching hits at $1 per million and 5-minute/1-hour cache writes at $12.50/$20.

The reversal is competitive. OpenAI's GPT-5.6 Sol ships similar benchmark scores at roughly a third of Fable's per-task cost, The Decoder adds, and the past eight days brought Meta's Muse Spark 1.1 at $1.25/$4.25, xAI's Grok 4.5 at $2/$6, and Moonshot's Kimi K3 at $3/$15, all pressing the frontier tier from below. Anthropic's move splits the difference: keep Fable bundled where usage is heaviest, meter it everywhere else.

Gemini 3.5 Pro has not launched, despite "released July 17" posts circulating

If you searched "Gemini 3.5 Pro release date" today, you may have hit posts titled "Google Gemini 3.5 Pro Launches" dated July 17. They are false. Google's own Gemini API release notes list only gemini-3.5-flash as generally available, and the Gemini API pricing page still shows no gemini-3.5-pro row, only Flash at $1.50/$9, Flash-Lite at $0.25/$1.50, and the 3.1-pro-preview. The "launched" posts, on domains like davidandgoliath.ai and faq.com.tw, cite no Google blog post, no model card, and no API ID.

The real status is unchanged from yesterday's lead: Bloomberg reported the delay July 16, and a Google spokesperson confirmed to Yahoo Tech that the company is "currently testing 3.5 Pro, an upgraded Flash model, and other models with partners," with no date. Mashable confirmed July 17 there is still no sign of the model. Gemini 3.5 Pro has now missed June, July 17, and beyond. Treat any "GA" claim as false unless gemini-3.5-pro appears in the public API docs.

SpaceXAI open-sourced Grok Build on July 15 and reset all usage limits

A quieter xAI move landed mid-week: on July 15, SpaceXAI posted on X that it had "open-sourced Grok Build and have reset usage limits for all users," publishing the Rust source for the Grok Build CLI and agent runtime under Apache 2.0, per Tony Reviews Things. The post framed the release as letting "anyone to support making a reliable and robust harness." Grok Build is the coding-agent harness that uses Grok 4.5 ($2/$6 per million tokens, 500K context) as its default model.

The usage-limit reset applies to every Grok Build user, though SpaceXAI did not specify whether the reset changes ceiling sizes or just clears existing consumption. The release sits alongside Elon Musk's broader open-sourcing push across his companies. Grok 4.5 itself remains closed-weight; only the agent harness is open. For developers evaluating Grok 4.5, the open CLI means you can now inspect and extend the harness layer that drives the model, and the reset gives a fresh quota to test with.

Tracking

Current prices, verified today against each vendor's official page

Prices are per 1 million tokens, standard tier, input then output, unless noted. Every row links to the official pricing or docs page, checked July 18.