Nanbeige4.3-33B Matches Qwen3.8 27B While Burning 44% More Tokens
11/06/2026 — 11/06, 03:19·1 sources (all)·2 reports (all)
Key takeaways
- Artificial Analysis has published a new breakdown showing Nanbeige4.3-33B tying Qwen3.8 27B (xhigh) with a score of 20 on the Intelligence I…
Background & analysis
On November 6, 2026, at 01:49 UTC, Artificial Analysis reported on X that Nanbeige, the foundation model research team at BOSS Zhipin, released Nanbeige4.3-33B, an open weights model. The release was announced via a post from the Artificial Analysis account, which stated that Nanbeige4.3-33B scored 20 on the Artificial Analysis Intelligence Index v5, matching Qwen3.8 27B (xhigh). According to the same announcement, Nanbeige4.3-33B showed stronger performance in terminal use, while lagging in knowledge-based tasks.
Also on November 6, 2026, at 01:49 UTC, Artificial Analysis published a token consumption comparison. Nanbeige4.3-33B averaged 158K output tokens per Intelligence Index task, versus 110K for Qwen3.8 27B, representing 44% more tokens despite both models scoring 20. This weighted average covers answer and reasoning tokens across the nine v5 evaluations, including agentic tool-use turns. No further updates on model revisions or official statements have been reported, and the public record currently stands at these two sets of figures.
AI-generated from 2 reports · updated 1 hour ago
Latest turnNanbeige4.3-33B used 158K output tokens per Intelligence Index task versus 110K for Qwen3.8 27B, 44% more even though both scored 20, according to Artificial Analysis. The weighted average covers answer and reasoning tokens across the nine v5 evaluations, including agentic tool-use turns.
Related AI tools
- Heat index
- 92
- All sources
- 1
- All reports
- 2
- First seen
- 3 hours ago
Reports on this story headlines open the original
Nanbeige4.3-33B used 158K output tokens per Intelligence Index task versus 110K for Qwen3.8 27B, 44% more even though both scored 20, according to Artificial Analysis. The weighted average covers answer and reasoning tokens across the nine v5 evaluations, including agentic tool-use turns.
X：Artificial Analysis (@artificialanlys)First-partyAI score 42Artificial Analysis reports that Nanbeige, the foundation model team at BOSS Zhipin, has released the open-weights Nanbeige4.3-33B. It matches Qwen3.8 27B (xhigh) at 20 on the Artificial Analysis Intelligence Index v5, with stronger terminal use but weaker knowledge, according to the evaluator.
X：Artificial Analysis (@artificialanlys)First-partyAI score 46
Other stories people are talking about
- Heat index 89NewQwen3.8 Flash-Next hits 250+ tok/s on a single RTX 50801 sources
- Heat index 58开发者深度实测 QoderCN 搭配 Qwen3.8-Flash 构建多场景应用1 sources
- Heat index 83Tutorial: single-GPU LoRA fine-tuning of Qwen3-VL-8B with Unsloth1 sources
- Heat index 43GPT-6 Astra directly rewrites Qwen weights to lift math accuracy1 sources
- Heat index 46Developer Releases RustingStrata, a Pure-Rust Qwen GGUF Inference Engine1 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion