Tied at 20, Nanbeige4.3-33B burns 158K output tokens per task, 44% more than Qwen
ArtificialAnlys · x · 2026-11-06
Across the nine Intelligence Index v5 evaluations, Nanbeige4.3-33B used 158K output tokens per task vs. 110K for Qwen3.8 27B—44% more despite both scoring 20. The weighted average includes answer and reasoning tokens, including agentic tool-use turns.
Related event: Nanbeige4.3-33B Shows Lower Hallucination but Higher Token Use(2 posts)→
More from Models
- Grok apologizes for unexpected Cursor charge as users question AI billing reliability — alexcovo_eth · 2026-11-06
- Ampersand Claims 92.2% on SWE-bench Pro HARD-51 at 70% Lower Cost — niloofar_mire · 2026-11-06
- LiteParse Integrates 1.3B MiniCPM-OCR, Beating Gemini Flash on ParseBench — llama_index · 2026-11-06
- GPT-6 meme: Astra does 40% less work when tasks are framed as safety research — repligate · 2026-11-06
- METR chart shows AI benchmark progress nearly doubled, Anthropic Model 4 hits 178.6 AECI — scaling01 · 2026-11-06
- Visa agentic commerce report: 22 shopping agents average 57.2% success, leaders hit 82% — omooretweets · 2026-11-06