Scan: Will Anthropic have the best AI model at the end of December 2026?

date	predicted	market price	confidence
2026-08-31	63%	70%	51%
2026-08-24	62%	70%	47%
2026-08-17	60%	68%	49%

1. Planner Fable 5.1 · Fermi decomposition + tool plan

Tools chosen: polymarket_direct resolution_source polymarket_related kalshi_related claude_news article_search gdelt_news wikipedia

Sub-questions (Fermi decomposition)

Which company's model currently holds rank #1 on the LMArena text leaderboard with style control OFF, and what is Anthropic's current best rank and score gap to #1?
Over the past 18-24 months, what fraction of time has an Anthropic model occupied the #1 rank on LMArena text (style control off), versus Google, OpenAI, and xAI?
What frontier model releases (Claude, Gemini, GPT, Grok, DeepSeek, etc.) are expected in H2 2026 and how close to December 31 are they likely to land?
Do Anthropic models systematically underperform on LMArena style-control-off relative to their benchmark performance (e.g., due to verbosity/formatting preferences), and has that pattern persisted?
What is the current Polymarket price for Anthropic and for the other companies in this market group, and do the sibling prices sum to roughly 100%?
Has LMArena changed its default leaderboard methodology (style control default, rank calculation, ties) in ways that could affect which company is shown at rank 1?

Planner reasoning

This is a Polymarket question resolving on the LMArena text leaderboard (style control off) on Dec 31, 2026. Key drivers: current leaderboard state and who holds #1, historical frequency with which Anthropic has topped LMArena (historically Google/OpenAI dominate, Anthropic rarely #1 with style control off), expected release cadence of Claude vs Gemini/GPT/Grok models through 2026, and the market's current price plus sibling markets for Google/OpenAI/xAI to sanity-check the implied distribution.

2. Research Parallel tool calls

tool	status	time	items	summary
polymarket_direct	OK	0.2s	1	## This Market's Polymarket Data Will Anthropic have the best AI model at the end of December 2026? - Current price (probability): 71.50% - 7-day price change: +3.00% - 30-day price change: +5.00% - Total volume: $156,764 (USD notional) - Price range: 60.00% - 72.50% - Data points: 90 days
resolution_source	OK	6.8s	1	already_resolved=unclear; matches_criteria=False; last_updated=Sep 13, 2026. The leaderboard (dated Sep 13, 2026) shows Anthropic's 'claude-fable-5-high' model in rank 1 with score 1506, and Anthropic models occupy ranks 1, 2, 3, 5, 6, 7 among the top entries, ahead of Meta, Google, OpenAI, and othe
polymarket_related	OK	0.7s	32	Polymarket search over 4 keyword(s), kept 32 matches. keyword 'best AI model end of 2026': 8 markets \| keyword 'best AI model': 8 markets \| keyword 'Anthropic': 8 markets \| keyword 'Chatbot Arena': 8 markets
kalshi_related	OK	3.4s	1	1 related markets / summaries. keyword 'best AI model': no matches \| keyword 'LMArena': no matches \| keyword 'Anthropic Claude': ok
claude_news	OK	26.6s	15	Based on available search results (as of mid-September 2026): - Current LMArena #1 (Anthropic-held): Multiple trackers indicate Anthropic's Claude currently sits atop the Text Arena leaderboard. One tracker notes "1510+ Frontier #1 Claude Opus 4.8 (AAII 61.4, coding & overall #1)" with the br
article_search	OK	0.9s	33	Fetched 33 unique articles across 4 queries (lookback=60d). 'LMArena leaderboard top model Claude Gemini GPT rank': 10 articles \| 'Anthropic Claude new model release LMArena score': 10 articles \| 'Gemini tops Chatbot Arena leaderboard': 10 articles \| 'OpenAI GPT model LMArena number one': 10 article
gdelt_news	OK	140.2s	20	GDELT: 20 articles across 3 queries (lookback=45d). 'LMArena leaderboard': 10 hits \| 'Anthropic Claude release': 10 hits \| 'Chatbot Arena ranking': error GDELT rate-limited after retries (429)
wikipedia	OK	0.4s	3	Fetched 3 Wikipedia entries (0 missing pages).

3. Evidence Brief Sonnet 5 · 6246 chars

# Event Will Anthropic (via LMArena text leaderboard rank #1, style-control off) hold the top-ranked AI model as of the Dec 31, 2026 12:00 PM ET check? # Outcomes to forecast - Yes (Anthropic ranked #1) - No (any other company ranked #1) # Kalshi market anchor No kalshi_direct data was returned for this ticker in the research pull. The closest Kalshi analog (kalshi_related) is unrelated to LMArena rank (IPO-race market). **Polymarket is used as the best available cross-market anchor**: current YES price 71.5%, up +3% (7d) and +5% (30d), range 60–72.5% over 90 days, volume $156,764. Treat this as directionally informative but not a literal Kalshi price. # Resolution source resolution_source tool fetched arena.ai (site rebranded from lmarena.ai in Jan 2026; redirects). As of Sep 13, 2026, the leaderboard shows Anthropic's "claude-fable-5-high" at rank #1 (score 1506), with Anthropic models occupying ranks 1,2,3,5,6,7 — a dominant sweep ahead of Meta, Google, OpenAI. Tool flags uncertainty because: (a) fetched from arena.ai not lmarena.ai exactly, and (b) snapshot is ~3.5 months before the actual Dec 31, 2026 check, so ranking could shift before resolution. # Sub-question answers 1. **Current #1 and gap** — Anthropic (Claude Fable 5 / Opus 4.8, naming inconsistent across sources) holds #1 as of Sep 2026 at ~1506-1525 Elo, but the gap to Gemini 3.1 Pro, GPT-5.5/5.6, and Grok 4.5 is narrow — within statistical noise (sub-25-50 Elo) [claude_news, resolution_source]. 2. **Historical share of #1** — Over the leaderboard's ~41-month history, OpenAI held #1 most often (15 months, 37%), Anthropic 9 months, Google 8 months; the crown changed hands 19 times [claude_news]. This suggests base-rate persistence of any single leader over a 3.5-month forward window is far from certain. 3. **Expected H2 2026 releases** — Google's Gemini 3.5/3.6 Pro flagship has been delayed with no announced timeline (only smaller Flash variants shipped) [article_search, claude_news]; Anthropic released Claude Opus 5/Fable 5 in July 2026; OpenAI shipped GPT-5.5/5.6; xAI's Grok 4.5 reportedly reached Opus-class performance at lower cost. No confirmed imminent Dec 2026 flagship releases identified. 4. **Style-control-off bias** — Not directly confirmed/denied in research; claude_news notes LMArena separates raw vs. style-controlled scores and that formatting/verbosity preferences can matter, but no specific evidence that Anthropic systematically underperforms style-off; current data shows Anthropic leading raw (style-off) rankings. 5. **Polymarket sibling prices** — Anthropic 71.5%, Google 9.5%, OpenAI 12.5%, Alibaba 0.75%, Z.ai 0.55%, xAI 2.95%, DeepSeek 0.65%, Moonshot 0.7%. Sum ≈ 98.6%, close to 100% (well-calibrated market group, minor takeout/rounding). 6. **Methodology changes** — Site rebranded lmarena.ai → arena.ai (Jan 2026); one source notes a July 12, 2026 re-baseline of scores to count only post-restoration votes. No evidence of change to rank/tie-break rules matching the market's stated methodology. # Key facts (high-confidence, factual) 1. [resolution_source] As of Sep 13, 2026, Anthropic occupies #1 (and multiple top-7 slots) on the arena.ai/lmarena.ai text leaderboard, style-control off. 2. [polymarket_direct] Polymarket Anthropic-YES = 71.5%, uptrending over 30 days. 3. [claude_news] Historically (41 months), OpenAI led #1 most often (37%), Anthropic 9 months, Google 8 months — leadership churns frequently (19 changes). 4. [article_search/claude_news] Google's Gemini 3.5 Pro flagship remains delayed as of Sep 2026, weakening near-term Google challenge. 5. [Wikipedia] Anthropic valued at $965B (May 2026), planning 2026 IPO; strong commercial/financial momentum coincides with model lead. 6. [claude_news] Open-weight models (DeepSeek V4.1 Pro, Kimi K3) are closing the gap (~55 Elo pts by June 2026), adding competitive noise but not yet threatening #1. # Cross-market signals - Kalshi related: No direct LMArena/best-model Kalshi market found; only tangential Anthropic IPO-race market (92% Anthropic wins IPO race) — not directly informative on model quality. - Polymarket: Anthropic 71.5% YES (rising); Google 9.5%, OpenAI 12.5%, xAI 2.95%, others <1% — sum ~98.6%, internally consistent. - Sportsbook implied: N/A (not applicable to this category). # Analyst opinions and speculation - claude_news synthesis: "four labs are within Elo-noise of each other," and rankings shift with each release; bottom line assessed Anthropic's continued dominance as "uncertain but plausible given current momentum." - Analysts note Google's delayed Gemini 3.5/3.6 Pro reduces near-term Google threat, but Google historically ships aggressively once ready, and DeepMind restructuring/talent departures reported (Sep 2026) could further delay a comeback. - xAI (Grok 4.5) and open-weight labs (DeepSeek, Kimi K3, Qwen 3.7/3.8) are narrowing gaps from below but not yet documented at #1. # Directional lean per outcome - **Yes (Anthropic)**: Currently #1 with multiple top slots (strong current status quo); Polymarket assigns 71.5% and rising; Google's flagship delayed; Anthropic has strong release cadence (Opus 4.6→4.7→4.8→5/Fable 5) and financial/commercial momentum. - **No (other company)**: Historical base rate favors frequent leadership churn (crown changed 19x in 41 months, Anthropic held #1 only 9/41 months historically); gap to Gemini/GPT/Grok is within statistical noise; ~3.5 months remain until resolution — ample time for at least one more flagship release cycle from competitors (esp. if Gemini 3.5/3.6 Pro finally ships). # Gaps / unknowns - No native Kalshi YES price was retrieved for this exact ticker — analysis leans on Polymarket as proxy. - Snapshot data is from mid-September 2026, ~3.5 months stale relative to the Dec 31 check date; no visibility into Oct–Dec 2026 model releases. - Uncertainty whether arena.ai fully equals the lmarena.ai resolution source specified in rules. # Calibration anchors - Polymarket YES price (proxy anchor): 71.5%, uptrending. - Historical base rate: any single company holds #1 in a given month only ~22-37% of the time (Anthropic specifically ~22% historically), though current momentum and delayed rivals favor continuation above base rate.

4. Forecasters 5 draws · Fable 5.1 + GPT-5.6 Sol

claude-fable-5-1

52% conf

Yes	67%
No	33%

The resolution source (arena.ai/lmarena text leaderboard, style-control off) shows Anthropic's Claude Fable 5 at #1 with Anthropic holding six of the top seven slots as of mid-September 2026, and the closest cross-market anchor (Polymarket) prices Anthropic YES at 71.5% and rising. Google's Gemini 3.5/3.6 Pro flagship remains delayed with DeepMind restructuring reported, weakening the most credible challenger over the remaining window. However, the leaderboard has changed leaders 19 times in 41 months, the Elo gap to Gemini 3.1 Pro, GPT-5.5/5.6, and Grok 4.5 is within statistical noise, and ~3.5 months leaves room for at least one competitor flagship cycle, so I shade modestly below the market. Net: status quo plus depth of Anthropic's top-rank sweep favors Yes, but churn risk justifies ~0.67.

claude-fable-5-1

52% conf

Yes	68%
No	32%

The resolution source (arena.ai, Sep 13, 2026) shows Anthropic at #1 with models occupying six of the top seven slots, and Polymarket prices Anthropic YES at 71.5% and rising, with Google at 9.5% and OpenAI at 12.5%. Status quo favors Anthropic, reinforced by Google's delayed Gemini 3.5/3.6 Pro flagship and Anthropic's rapid release cadence. However, ~3.5 months remain before the Dec 31 check, the leaderboard crown has changed hands 19 times in 41 months, and the brief notes the gap to Gemini/GPT/Grok is within Elo noise, so a single competitor flagship release could flip rank #1. I shade slightly below the Polymarket anchor to reflect leadership churn and the staleness of the snapshot relative to the resolution date.

claude-fable-5-1

52% conf

Yes	67%
No	33%

As of the mid-Sep 2026 resolution-source snapshot, Anthropic holds #1 on the arena.ai text leaderboard (style-control off) and occupies six of the top seven slots, a strong status-quo position with ~3.5 months to the Dec 31 check. Polymarket prices Anthropic YES at 71.5% and rising, with rivals summing consistently (Google 9.5%, OpenAI 12.5%, xAI ~3%), and Google's Gemini 3.5/3.6 Pro flagship remains delayed. However, the leaderboard has historically churned frequently (19 leadership changes in 41 months, Anthropic #1 only ~22% of months) and top-lab gaps are within Elo noise, so a single competitor flagship release in Q4 could flip the rank. I shade slightly below the Polymarket anchor to reflect the stale snapshot and churn base rate.

gpt-5.6-sol

68% conf

Yes	69%
No	31%

The best available market anchor is Polymarket at 71.5% YES, and Anthropic currently holds rank #1 while occupying six of the top seven leaderboard positions. Delays to Google's next flagship and the absence of confirmed imminent competitor releases support persistence through December. However, the lead is within roughly 25–50 Elo of several rivals, while the crown changed hands 19 times in 41 months, making a change over the remaining 3.5 months material. I therefore shade slightly below the market anchor to 69% because the stale snapshot and frequent historical churn outweigh some of Anthropic's current dominance.

gpt-5.6-sol

68% conf

Yes	68%
No	32%

Anthropic currently holds rank #1 and six of the top seven positions, so its depth points toward Yes even though the lead over rival labs is within Elo noise. Historical evidence points toward No more than the current snapshot alone would suggest, because leadership changed 19 times in 41 months and roughly 3.5 months remain. Delays to Google's next flagship and the absence of a confirmed imminent rival release support persistence, while style-control-off currently shows no documented disadvantage for Anthropic. I therefore forecast slightly below the 71.5% Polymarket proxy anchor, pricing the substantial release-cycle and ranking-churn risk at 32%.

6. Final plain mean of 5 draws

68% (58% confidence)

Yes: 68% No: 32%

Mean of 5 draws (0.67, 0.68, 0.67, 0.69, 0.68). The resolution source (arena.ai/lmarena text leaderboard, style-control off) shows Anthropic's Claude Fable 5 at #1 with Anthropic holding six of the top seven slots as of mid-September 2026, and the closest cross-market anchor (Polymarket) prices Anthropic YES at 71.5% and rising. Google's Gemini 3.5/3.6 Pro flagship remains delayed with DeepMind restructuring reported, weakening the most credible challenger over the remaining window. However, the leaderboard has changed leaders 19 times in 41 months, the Elo gap to Gemini 3.1 Pro, GPT-5.5/5.6, and Grok 4.5 is within statistical noise, and ~3.5 months leaves room for at least one competitor flagship cycle, so I shade modestly below the market. Net: status quo plus depth of Anthropic's top-rank sweep favors Yes, but churn risk justifies ~0.67.

Pipeline Timing

Total pipeline time: 223.0s

Per-tool research timings shown in the Research section above.