DeepSWE long-horizon coding leaderboard: Gemini 4 Argon third at 77.9%

Stats Wire @StatsWireOriginally posted 2026.10.15

Open original on X

While everyone is arguing over OpenAI and Google, Meta and Anthropic are quietly eating their lunch. Look at Long-Horizon Coding (DeepSWE) right now: • Meta Muse Wonder Full → 79.0% (#1) • Meta Muse Wonder → 78.8% (#2) • Gemini 4 Argon → 77.9% • Claude Opus 5.5 → 74.2% • GPT-6 Astra → 74.1% • Claude Fable 5.1 → 67.4% And over the last 12 months of AI web traffic: • ChatGPT → Dropped from 72% → 56% • Gemini → Nearly doubled to 25% • Claude → Surged 4.2× (2.3% → 9.6%), now larger than DeepSeek, Grok, Copilot & Perplexity combined. The 1-company monopoly is officially over.