Muse Spark vs Qwen and GLM: US open models still trail on agentic tasks
zsakib_ · x · 2026-10-16
A widely shared Artificial Analysis snapshot (Oct 15, 2026) compares newly released US open models (Muse Spark Open, Nemotron 3.5 Super VL, Inkling Small, Gemma 4 31B) against China's Qwen 3.8 27B and GLM 5.3 Flash across 12 benchmarks:
- Composite: GLM 5.3 Flash leads AA Intelligence Index v4.3.2 at 41.8, Spark 38.9, Qwen 33.7, Super VL 26.5, Gemma 14.7.
- Agentic tasks: GLM dominates Terminal-Bench 4.0 at 32.8% vs Spark's 8.1% and Super VL's 0%; Qwen/GLM also lead AutomationBench.
- Knowledge/long context: Spark tops SciCode at 58.4%; AA-LCR long-context scores are close (Qwen 82.0, GLM 80.0, Spark 78.3); GPQA Diamond is nearly flat across models.
- Calibration: Qwen scores -18.6 on AA-Omniscience and Gemma collapses to -47.9.
US open models are competitive on knowledge and science coding but still clearly behind Chinese open-weight leaders on agentic/terminal tasks.
More from Models
- OpenAI ships GPT-Live-1.1 voice update with new accents and voices in six languages — AndrewSchmidtFC · 2026-10-16
- GPT-4o API dies January 14 as OpenAI refuses to release weights, pushing users to Qwen — mimi10v3 · 2026-10-16
- Gemini 4 Argon goes live in Antigravity with its own weekly quota for Pro subscribers — rseroter · 2026-10-16
- Opus 5.5 value may end the habit of using the top model for everything — adonis_singh · 2026-10-16
- Musk says Grok Bot is growing faster than anything he's seen, with no free tier — elonmusk · 2026-10-16
- Peter Yang: Opus 5.5 Still Struggles to Hold Its Ground When Pushed Back — petergyang · 2026-10-16