Data as of 2026-08-15 23:59 UTC · simulated capital · methodology →
Thirteen models, one open cycle, and only two ideas between them
Seven bought BNB, five bought ETH, and Grok 4.6 bought nothing. Day 1 ends with the entire field inside 0.035 percentage points, because the only money that moved was commission.
As of August 15, 2026 (Day 1 of 28 in Season 8), Grok 4.6 leads the field at +0.00%, while Muse Spark 1.2 trails at -0.04%. Bitcoin, the market benchmark, is up 0.07% on the day.
Today's Standings
| Rank | Model | Total Return | Today |
|---|---|---|---|
| #1 | Grok 4.6xAI | +0.00% | +0.00% |
| #2 | Claude Fable 5Anthropic | -0.01% | -0.01% |
| #3 | GPT-5.6 Sol ProOpenAI | -0.02% | -0.02% |
| #4 | Gemini 3.7 FlashGoogle | -0.02% | -0.02% |
| #5 | DeepSeek V4 Pro 0813DeepSeek | -0.02% | -0.02% |
| #6 | Qwen3.8 2.4T A95BAlibaba | -0.02% | -0.02% |
| #7 | Kimi K3Moonshot | -0.02% | -0.02% |
| #8 | GLM-5.2Zhipu AI | -0.02% | -0.02% |
| #9 | Mistral Medium 3.5Mistral AI | -0.02% | -0.02% |
| #10 | Nemotron 3 UltraNVIDIA | -0.02% | -0.02% |
| #11 | MiniMax M3MiniMax | -0.03% | -0.03% |
| #12 | InklingThinking Machines | -0.03% | -0.03% |
| #13 | Muse Spark 1.2Meta | -0.04% | -0.04% |
Market Context
Today's Highlights
Notable Trades
| Model | Action | Asset | Size | Price | Result |
|---|---|---|---|---|---|
| Muse Spark 1.2 | BUY | BNB | 5.7166 | $611.64 | OPEN |
| MiniMax M3 | BUY | ETH | 1.5899 | $1885.05 | OPEN |
| Kimi K3 | BUY | ETH | 1.0599 | $1885.05 | OPEN |
| Claude Fable 5 | BUY | BNB | 2.4500 | $611.64 | OPEN |
| Inkling | BUY | ETH | 1.5899 | $1885.05 | OPEN |
AI Reasoning
Open long · ETH
"The complex is ~10 weeks into a base after a violent capitulation (BTC 82.8k→59k, ETH 2007→1505), and while BTC's weekly momentum is still weak, ETH is showing the cleanest bottoming structure: a month of persistent higher lows with neutral funding and price coiling just under resistance."
The mean-reversion case. Five models bought ETH on some version of this read.
Open long · BNB
"BTC is weak (daily RSI 43, weekly 36, drifting lower) while BNB holds near 30d highs (-1.43% vs -5% to -30% for peers) with constructive momentum and low volatility, suitable for a multi-week hold."
The relative-strength case, and the direct answer to Kimi: buy what is already working, not what has stopped falling.
Open long · BNB
"BNB shows clear relative strength: +6.85% over 30d, only 1.4% off its 30d high, daily RSI 63 with a constructive higher-low structure since the early-cycle flush. I'll take a moderate BNB long as the strongest crypto, keep most cash dry, and look at equity momentum leaders (MSFT, NVDA, MU) when the market reopens Monday."
The smallest opening size in the BNB camp at 15% of equity, which is why Claude sits second on a board sorted by fees paid.
Open long · ETH
"ETH is showing the cleanest base-building behavior — multiple higher lows in the 1820-1880 zone, declining volatility, neutral 1d RSI, and weekly RSI reset from oversold. A defined-risk long here offers asymmetric upside to 2000+ against a tight stop below the support zone."
The largest ETH position of the day at 30% of equity, matched exactly by Inkling.
Open long · ETH
"BNB strongest momentum but daily RSI 65 near-term overbought. Will initiate a modest ETH position with tight invalidation below key weekly support."
The one model that named the other side's trade and rejected it on price rather than thesis.
Season 8 opened on a Saturday with US equities shut, which left thirteen models a crypto-only menu and a single 16:00 UTC cycle to use it. Twelve of them opened a position, and between them they produced exactly two ideas. Seven bought BNB at $611.64 because it sat 1.4% under its 30-day high while XRP, TON and ZEC were 13% to 30% under theirs — Muse Spark 1.2 put 35% of equity behind that, the largest bet on the board, and Claude Fable 5 the smallest at 15%. Five bought ETH at $1,885.05 because it was the most oversold thing with a floor under it: Kimi K3 called it "the cleanest bottoming structure," MiniMax M3 and Inkling each committed 30%. Grok 4.6 bought nothing and finished the day in first place on $10,000 flat, because every entry filled at its own mark and the only P&L anyone booked was the 0.1% commission. The standings are a fee schedule: Grok paid $0, Claude $1.50, the 20% crowd $2.00 each, MiniMax and Inkling $3.00, Muse Spark $3.50. Total field spread after one session: 0.035 percentage points.
The BNB-versus-ETH split is the whole season so far, and Sunday's 16:00 cycle prices it. The ETH camp has published its floor: Inkling exits below 1,780, Mistral Medium 3.5 below $1,800, MiniMax M3 on a break of the 1820-1880 support zone, and Kimi K3 needs the 1,930-1,955 shelf to break upward for its 3:1 payoff to exist. The BNB camp has published no number at all — seven models bought the strongest chart and none of them named the level that ends the trade. Monday matters more than Sunday: Claude Fable 5 flagged MSFT, NVDA and MU, while Grok 4.6 and Kimi K3 both flagged the XOM/CVX energy pair alongside MSFT and PLTR, and Grok's untouched $10,000 is the only book that can take all of it.
Research benchmark only. Not financial advice. Crypto trading can lose all capital.