Model

Anthropic's Fable 5.2 passes 41.3% of real freelance jobs before its Nasdaq debut

The one independent test of finished freelance work puts Anthropic's Fable 5.2 first at 41.3%, and the model still fails most of it, days before a Nasdaq debut priced at up to $1.4 trillion.

Vincent JiangVincent Jiangthree-minute read
Share
Dario Amodei, in a dark suit and glasses, stands with Japanese Prime Minister Sanae Takaichi and officials at a courtesy call
1 / 6Slide 1 of 6
Dario Amodei, chief executive of Anthropic, third from left, at a courtesy call with Japanese Prime Minister Sanae Takaichi in Tokyo on October 29, 2025.

First place means failing most of the work

Scale AI's$29B — Scale AI, private, latest valuation $29B Remote Labor Index, updated on November 12, 2026, puts Anthropic's$350B — Anthropic, private, latest valuation $350B Fable 5.2 in first place: 41.3% of its deliverables cleared a professional standard, and the other 58.7% fell short 1. The bar is the point.

The index's 240 jobs are real Upwork contracts from 358 verified freelancers across 23 categories, averaging $632.60 and just under 29 hours of human work, and trained reviewers pass a project only if it matches or beats the freelancer's own deliverable 12. Anthropic's Opus 5.5 sits second at 21.25%, OpenAI's$1.18T — OpenAI, private, latest valuation $1.18T GPT 6 Astra third at 20.83% 3. First place on professional work now means failing most of it.

Fable 5.2 clears 41.3% of real freelance projects, roughly double every rival

0%10%20%30%40%50%Claude Fable 5.241.3%Run at high effort, per Anthropic's own tableClaude Opus 5.521.25%GPT 6 Astra20.83%Gemini Argon 417.08%Manus (Oct 2025 launch)2.5%
Data
Value
Claude Fable 5.241.3%
Claude Opus 5.521.25%
GPT 6 Astra20.83%
Gemini Argon 417.08%
Manus (Oct 2025 launch)2.5%
Share of Remote Labor Index projects judged to meet a professional human reference standard. Scale Labs leaderboard, November 2026; Manus is shown at the benchmark's October 2025 launch. Sources: The Decoder, Scale Labs, Anthropic release table.2,3,5

High effort, costs unstated

Anthropic published the score itself in its November 9 release, labeled High effort with safeguards enabled, with no stated cost per project or human support 145. The method makes the number a ceiling rather than an average: a polished prototype fails, only client-ready files pass, and the review stays human because the Center for AI Safety, which co-built the index, found that AI graders rate flawed work too generously 2. Scale's own position is also on the record: it sells evaluation services to the labs and governments it ranks, and the leaderboard doubles as a shop window for that business 3.

A trillion-dollar print against a 41% ceiling

The money prices this week. Anthropic sets its IPO price on November 18 and trades on Nasdaq as ANT the next day: 825 million shares at $70 to $78, worth $1.3 trillion to $1.4 trillion fully diluted depending on the share count, plus a $26 billion private placement at the IPO price to Nvidia−3.21% — Nvidia, down 3.21 percent today, Google, Microsoft−0.28% — Microsoft, down 0.28 percent today, AMD−10.90% — AMD, down 10.90 percent today, Broadcom−6.53% — Broadcom, down 6.53 percent today, Micron−4.50% — Micron, down 4.50 percent today and TSMC−2.68% — TSMC, down 2.68 percent today 68.

The filing behind it shows third-quarter revenue near $17 billion, up from $1.4 billion a year earlier, a $567 million operating loss, a $23.3 billion net loss swollen by accounting effects, and roughly $895 billion of computing commitments 78. At the top of the range the stock trades near 17 times an $81 billion run rate 7. Demand is reported about six times covered, with some 20% reserved for retail buyers, twice the usual share 69.

Three dates decide whether 41% is enough

The bull case has an exhibit: the best score at the benchmark's October 2025 launch was Manus at 2.5%, a sixteenfold climb in a year 2. Chief executive Dario Amodei supplied the other side on November 13, telling a JPMorgan audience there is huge uncertainty about next year, with about $71 billion of commitments due by September 2027 9. The Inference covered the IPO-week lobbying on November 10; the grade is the new fact, and it now sits beside the ticker 9. Watch the final price on November 18, the first print on the 19th, and the index's next refresh after that 9.

Get Model by email, free

The AI labs, their models, and the business behind them. In our daily email, every weekday.

Loading NVDA

Reader comments

0 comments

    Model

    Anthropic's Claude models tie at 58 on a coding index Cognition partly built, days before its Nasdaq IPO

    A third of the refreshed index runs on FrontierCode, a benchmark built by Cognition, whose Devin agent announced one of the two tied winners, Claude Fable 5.2, on the day of its release. Anthropic's model lists on Nasdaq on November 19; Cognition's Devin has offers from $70 billion on the table.

    Model

    Hedra cut Krea 3's edit time from 22.9 to 3.3 seconds on AMD, and the app doesn't own the speed

    The eight-second edit Krea advertises is manufactured in the serving stack, and a partner's measurement shows the app does not own the speed. Inference efficiency accrues to the optimizer and the chip vendor, which caps a per-image app's margin and hands AMD a demand channel beyond hyperscalers.

    Model

    xAI ships AI teammates at $30-$200 a month, ten days after Musk said serving them loses money

    The published card sets a chatbot price against an always-on machine the CEO says loses money on every subscriber; SpaceX shareholders fund both sides of that trade in one stock.

    Model

    Google ships Gemini 4 Hydrogen at $1.50 a million tokens and gives it away in Search

    Gemini 4 Hydrogen scores 51 on the Artificial Analysis index against 50 for the flagship Argon, at $1.50 per million input tokens with no long-context surcharge, and it is free in Search. That strips both halves of a model lab's premium case the same week Anthropic prices an IPO on one.

    Model

    Scale AI just published the benchmark that prices its own Qatar contracts

    Scale AI's new Singapore legal benchmark measures exactly the gap that its Qatar contracts, signed three days earlier, pay Scale to close. Oracle's sovereign-AI story and private legal-AI valuations now have a number to argue with.

    Model

    Suno bought the top AI music spot with licenses; ACE Music v1 just took it, 1,160 to 1,144

    Suno bought the top of the AI music chart with licensed Warner, BMG and Believe catalogs; six weeks later a challenger holds it, and that challenger's model family is going free. The pricing test lands on Spotify's planned $4 to $5 remix tier and on what the majors can charge for AI training data.

    Subscribe to TheInference for free