--- title: "Anthropic's Claude models tie at 58 on a coding index Cognition partly built, days before its Nasdaq IPO" description: "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." publisher: "The Inference" section: "Model" published: 2026-11-15T15:29:32.124Z modified: 2026-11-15T15:29:32.124Z canonical: https://theinference.org/article/anthropic-s-claude-models-tie-at-58-on-a-coding-index-cognition-partly-built-days-before-its-nasdaq-ipo language: en keywords: "AI Models, Benchmarks, AI Coding, IPOs" --- # 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. The yardstick in coding agents now sits inside a seller's shop. On November 13, 2026, Artificial Analysis, an independent model-benchmarking service, [refreshed its Coding Agent Index](https://cryptobriefing.com/artificial-analysis-coding-agent-index-v2-1/), and the top composite fell from 61.7 to 58, with [Anthropic](https://theinference.org/markets/companies/anthropic)'s Claude Fable 5.2 and Claude Opus 5.5 tied at first [1]. A third of that score runs on FrontierCode, a task suite built by [Cognition](https://theinference.org/markets/companies/cognition), the startup behind the Devin coding agent [1]. Cognition announced [Fable 5.2 inside Devin Desktop and Devin CLI](https://www.remio.ai/post/claude-fable-5-2-comes-to-devin-desktop-and-cli-with-a-new-coding-benchmark-lead) on November 9, the day Anthropic released it [1][2]. Anthropic's model lists on Nasdaq this week [3]; Cognition's Devin, the harness that runs it, has fielded offers that could close a round before the year ends [4]. *Chart: **The top score fell from 61.7 to 58 when the index changed its tests** Column chart of composite scores on the Artificial Analysis Coding Agent Index: Claude Opus 5.5 scored 61.7 on version 2.0 and 58 on version 2.1, where Claude Fable 5.2 also scored 58.* *Composite scores on the Artificial Analysis Coding Agent Index, which scores model-and-harness pairs; version 2.1, released November 13, 2026, refreshed the benchmark suite. Reported figures, in index points. [1]* ## Cognition owns the harness and the grading, not the champion The champion is Anthropic's. Fable 5.2 leads the new index and wins Cognition's own test: 49.1 on FrontierCode 2.0 at the highest reasoning setting [2]. The index runs version 1.1 of that suite; the 49.1 was scored on version 2.0, a different edition of Cognition's own benchmark [1][2]. Cognition's own small model is the discount instrument. SWE-2 Flash, [released on November 4 and sold inside Devin at a fifth of SWE-2's price](https://www.aiaffairs.com/technology/cognition-swe-2-flash-coding-model-price-speed/), wins one of eight rows on Cognition's launch table ([5][6]; [our earlier take](https://theinference.org/article/cognition-s-4-8x-benchmark-proves-throughput-not-cheaper-agent-work)). What Cognition holds is the harness, the scaffolding around a model where a task gets its price; Devin's invoice folds model, harness and serving into one bill [6]. Devin's business is running someone else's frontier model cheaply, and grading the result. ## A leadership claim depends on which version of the test ran FrontierCode grades an agent's work on whether a repository maintainer would merge it, closer to how a team judges code than any correctness test [1]. But Cognition builds and versions the suite, and Artificial Analysis chose to swap version 1.1 into this refresh [1]. Opus 5.5 fell from 61.7 to 58 on an unchanged model and a changed test; that is what a new grader does to an old score [1]. The chain underneath is short. A frontier model writes mergeable code; the harness turns it into a task with a price; routine engineering work then moves off people's calendars. The first two links are priced this month; the third is projection. ## Anthropic prices this week; Cognition's offers run to year-end Anthropic begins trading on Nasdaq on November 19, 2026, at a valuation of up to about $1.4 trillion [3]. That is below the range investors contemplated a few weeks ago; the halo around AI, in one investor's words, "isn't as formidable" as it was [7]. At the top of the range the shares trade at roughly 17 times the $81 billion annualized run-rate in the filing [3]. ## Cognition's offers run at five times Anthropic's run-rate multiple Cognition is [set to raise about $1 billion at roughly $47 billion](https://theinference.org/article/cognition-s-president-says-smarter-models-no-longer-win-as-robinhood-s-fund-buys-in-at-48-billion), per Bloomberg, on a run-rate above $900 million, with no full-year figures yet published [8]. It has separately received offers between $70 billion and $80 billion and may close a round before the end of the year [4]. Per dollar of run-rate, Devin's harness is the expensive asset and the model it runs is the cheap one. Anthropic's opening trade prices the model on November 19; a Cognition term sheet, if it lands this year, prices the grader [4]. *Chart: **Cognition's offers price each dollar of run-rate about five times higher than Anthropic's IPO** Bar chart of valuation as a multiple of annualized run-rate: Anthropic at 17.3 times, Cognition at 52.2 times in its $47 billion round, 77.8 times at a $70 billion offer and 88.9 times at an $80 billion offer.* *Valuation divided by annualized run-rate, in multiples. Anthropic: $1.4 trillion at the top of the IPO range against the $81 billion annualized run-rate in its filing. Cognition: a ~$47 billion round on a run-rate above $900 million. The $70 billion and $80 billion offers are indicated terms, not transacted prices. Multiples calculated from reported figures. [3][4][8]* ## Takeaway The top of the coding index fell because the test changed, not the models, and the two companies whose assets it prices, Anthropic's model and Cognition's harness, hit the market at very different multiples of run-rate. ## Sources 1. [Crypto Briefing, "Artificial Analysis Coding Agent Index v2.1 refreshes its tests, and two Claude models tie at the top", 13 November 2026](https://cryptobriefing.com/artificial-analysis-coding-agent-index-v2-1/) 2. [Remio, "Claude Fable 5.2 Comes to Devin Desktop and CLI With a New Coding Benchmark Lead", 10 November 2026](https://www.remio.ai/post/claude-fable-5-2-comes-to-devin-desktop-and-cli-with-a-new-coding-benchmark-lead) 3. [CoinCentral, "Anthropic IPO 2026: Should You Buy the Stock When It Goes Public?", 15 November 2026](https://coincentral.com/anthropic-ipo-2026-should-you-buy-the-stock-when-it-goes-public/) 4. [Surveillance Ghana, "Devin Maker Cognition Eyes Eighty Billion Dollar Valuation Amid AI Coding Boom", 9 November 2026](https://www.surveillanceghana.com/devin-maker-cognition-eyes-eighty-billion-dollar-valuation-amid-ai-coding-boom/) 5. [AI Affairs, "Cognition releases SWE-2 Flash at a fifth of SWE-2's price", 6 November 2026](https://www.aiaffairs.com/technology/cognition-swe-2-flash-coding-model-price-speed/) 6. [OrcaRouter, "Cognition SWE-2 Flash: The Discount Is Real, the Base Model Is the News", 5 November 2026](https://www.orcarouter.ai/blog/cognition-swe-2-flash-release) 7. [Yahoo Finance, "Anthropic IPO to test Wall Street's faith in AI as skepticism mounts", 14 November 2026](https://finance.yahoo.com/markets/stocks/article/anthropic-ipo-to-test-wall-streets-faith-in-ai-as-skepticism-mounts-180007104.html) 8. [Paul's AI, "Cognition, the startup behind AI engineering agent Devin, is set to raise ~$1 billion at a ~$47 billion valuation", 9 November 2026](https://pauls-ai.beehiiv.com/p/cognition-the-startup-behind-ai-engineering-agent-devin-is-set-to-raise-1-billion-at-a-47-bill) --- Anthropic's Claude models tie at 58 on a coding index Cognition partly built, days before its Nasdaq IPO — The Inference. Canonical: https://theinference.org/article/anthropic-s-claude-models-tie-at-58-on-a-coding-index-cognition-partly-built-days-before-its-nasdaq-ipo