Quantreno Research
AI Model Milestones
Claude Opus 5 holds the LMArena top spot into Aug 31; Gemini 3.5 Pro's delays make a late upset unlikely.
as of Sat 10:45 AM UTC · 5d agoStale· re-researched Sat after automated verification held the prior view· next: October rungs close — iPhone 18 plus GPT-6/Astra/Mythos ladder — Oct 1 (confirmed)What's moving. Anthropic still owns the board: Claude Opus 5 (shipped Jul 24) leads LMArena's preference leaderboard at 12.19% win-share, with Claude Fable 5 right behind at 12.01%, and Opus 5 took the WebDev Arena top spot (1,691 Elo) from Fable 5 as of mid-August, per Tech Insider. Google's live flagship remains Gemini 3.1 Pro, which trails Opus 5 badly on coding — 54.2% vs 64.3% on SWE-bench Pro per Vellum-cited figures. The swing factor is Gemini 3.5 Pro: announced in May, promised for June, repeatedly delayed on reported performance issues, and spotted A/B testing on LMArena in early August (NPowerUser) — but prediction-market odds of a mid-August launch were cut sharply on the continued slippage (Pluang).
Why now. The final Gemini 3.5 Pro release rung closes Aug 31, and Sept 1 opens the GPT-6/Astra/Mythos/Opus 6/Grok 5 ladder (OddsShopper's resolution calendar).
Variant perception: none — this is the consensus read. Performance-driven delays argue against a chart-topping Gemini debut inside the window, and the evidence shows no underpriced tail.
Evidence gaps. The Kalshi scan surfaced only corporate OpenAI/Anthropic series, all closing beyond 120 days; no live leaderboard-market quotes were pulled, and X searches returned nothing on stealth checkpoints.
Sources
- Kalshi · KXUSOPENAIANTH: 0 open markets close inside 120 days — 3 open beyond the window (earliest close 2027-01-01); KXCOMPANYACTIONANTH: 0 open markets close inside 120 days — 1 open beyond the window (earliest close 2027-01-01); KXOAIANTH: 0 open markets close inside 120 days — 2 open beyond the window (earliest close 2040-01-01) · true as of Sat 10:45 AM UTC · 5d ago · fetched Aug 29, 10:45 AM
- Web · Claude vs ChatGPT vs Gemini 2026: 88% SWE-Bench, $2 API · true as of Jun 26, 12:17 PM UTC · 2mo ago · fetched Aug 29, 10:45 AM
- Web · Claude Opus 5 vs GPT-5.6 Sol vs Qwen 3.8 Max: 4x Price Gap [2026] · true as of Aug 22, 10:00 AM UTC · 12d ago · fetched Aug 29, 10:45 AM
- Web · Claude vs ChatGPT vs Gemini: Lead Shrinks to 54% [2026] · true as of Aug 8, 10:00 AM UTC · 26d ago · fetched Aug 29, 10:45 AM
- Web · Gemini 3.6 Flash API Tutorial: 12 Steps [2026] · true as of Aug 25, 10:09 AM UTC · 9d ago · fetched Aug 29, 10:45 AM
- Web · Tech Release Date Odds: The Hype Is 0-for-16 on Kalshi · true as of Aug 15, 10:00 AM UTC · 19d ago · fetched Aug 29, 10:45 AM
- Web · Claude Opus 5 vs Grok 4.6 vs Gemini 3.1 Pro: $19 Gap [2026] · true as of Aug 26, 10:00 AM UTC · 8d ago · fetched Aug 29, 10:45 AM
- Web · Google's Gemini 3.5 Pro Launch Is Months Late - Technology Org · true as of Jul 17, 10:07 AM UTC · 1mo ago · fetched Aug 29, 10:46 AM
- Web · Odds drop sharply on imminent Google Gemini Pro... | Pluang · true as of Aug 15, 10:00 AM UTC · 19d ago · fetched Aug 29, 10:46 AM
- Web · Gemini 3.5 Pro Spotted on LM Arena: Launch Imminent? - NPowerUser · true as of Aug 2, 12:12 PM UTC · 1mo ago · fetched Aug 29, 10:46 AM
- Web · Claude Sonnet 5: 63.2% SWE-bench, $2/M Tokens [2026] – Tech Insider Canada · true as of Jul 30, 10:00 AM UTC · 1mo ago · fetched Aug 29, 10:46 AM
- Web · Claude vs Gemini 2026: 82.1% vs 63.8% SWE-bench [Tested] · true as of Apr 11, 5:37 AM UTC · 4mo ago · fetched Aug 29, 10:46 AM
The research is step one
Quantreno is an AI portfolio manager for your own account. The same desk that researched this brief can shape the idea into a sized, risk-checked draft trade — and nothing places until you approve it. Start free in early access: simulated money, zero dollars at risk — stock ideas place on Alpaca paper, event ideas on Kalshi paper; live trading opens at launch.