← Back to Hello, AI
2 min

Fable 5.2 and Qwen4-Max Ship Unrated. Argon Leads.

Claude Fable 5.2 and Qwen4-Max ship without a November 11 text score. Gemini 4 Argon has one, and it now leads the tracked set.

Three cards on helloai moved this week, and only one earned a text score. Claude Fable 5.2 replaced Claude Fable 5.1 on November 9 at API id claude-fable-5-2. Anthropic lists $8 per million input tokens and $40 per million output tokens, with cache reads at $0.20 and a 1 million token context. That replaces the Fable 5.1 list of $10 and $50. Qwen4-Max went generally available on November 16 as qwen4-max. QwenCloud's Singapore page lists $1.50 input and $4.50 output per million tokens and a 1 million token context, replacing Qwen3.8-Max at $2 and $6.

Gemini 4 Argon is the card with its own score. Google's pricing page lists gemini-4-argon on the paid tier at $2 input and $10 output per million tokens through November 30, 2026, then $4 and $20, with no free tier and a 1,048,576 token context. Scores below are the November 11 style-control text snapshot in the public lmarena-ai dataset, rounded half up. arena.ai returned a Cloudflare challenge, so the rendered page was not read. Gemini 4 Argon (High) is 1522, bounds 1516 to 1527, on 20,779 votes. Claude Opus 5.5 (High) is 1507 on 15,793 votes. Muse Spark 1.3 (Max) is 1497 on 26,806 votes. Grok 4.7 (xHigh) is 1443 on 22,219 votes.

Claude Fable 5.2 and Qwen4-Max have no row on that text board. The Fable card shows Claude Fable 5.1's November 11 score of 1502 and labels it a predecessor. The Qwen card does the same with Qwen3.8-Max at 1484. Neither model is ranked. Gemini 4 Argon is ranked, and it is the highest rated model in the set, so Overall Preference now points at it. Gemini 4 Carbon (High) is rank 1 in the same file at 1526 on 4,273 votes. That file has no preliminary column. Rendered-board write-ups still mark Carbon preliminary, so Carbon stays off the roster.

Code Arena is a different board. The dataset's WebDev overall snapshot is dated November 15. Claude Fable 5.2 (xHigh) leads it at 1826 on 1,445 votes. Claude Opus 5.5 (Max) is next at 1816 on 4,399 votes. Qwen4-Max (xHigh) is 1750 on 1,814 votes, and Grok 4.7 (xHigh) is 1640 on 8,488 votes. A WebDev lead with no text row stays out of the recommendation ranking. Coding & Engineering still names Claude Opus 5.5. Muse Spark 1.3 remains the cheapest tracked API at $1.25 and $4.25.

The cap stays at six. Muse Wonder 1.0 is public at $4 and $20 per million tokens with a 1 million token context, and Muse Wonder (Max) is 1511 on 12,696 votes, a second snapshot after October 15. It stays off so Muse Spark 1.3 keeps the cheap rate. GPT-6.1 Sol (Max) at 1481, GLM-5.3 (Max) at 1482, GPT-6 Astra (Max) at 1474, and Kimi K3 (Max) at 1487 clear the text floor and wait on a cap decision. Grok 4.7 stays at $2 and $6 below 200K prompt tokens and $4 and $12 above, with a 500K context. December 1 is when Argon's list price doubles.

← More from Hello, AI