[Monitor] Frontier update — week of 2026-08-31

[github-actions]

🔭 Frontier model & benchmark monitor — 2026-08-31

Auto-generated by monitor_models.py · lookback: 7 days

🤖 New open-weight model releases (Hugging Face Hub)

• Qwen/Qwen3.8-Flash-Next-FP8 — Qwen · ⬇️ 76,935 · ❤️ 164
  • Tags: transformers, safetensors, qwen4_exp, image-text-to-text, conversational
• Qwen/Qwen3.8-Flash-Next — Qwen · ⬇️ 121,976 · ❤️ 4411
  • Tags: transformers, safetensors, qwen4_exp, image-text-to-text, conversational
• microsoft/mattergen — microsoft · ⬇️ 0 · ❤️ 53
  • Tags: materials-science, generative-ai, materials-discovery, license:mit, region:us
• nvidia/GLM-5.2-NVFP4 — nvidia · ⬇️ 1,186,756 · ❤️ 319
  • Tags: Model Optimizer, safetensors, glm_moe_dsa, nvidia, ModelOpt
• nvidia/Qwen3.6-35B-A3B-NVFP4 — nvidia · ⬇️ 11,202,552 · ❤️ 577
  • Tags: Model Optimizer, safetensors, qwen3_5_moe, nvidia, ModelOpt
• nvidia/Kimi-K2-Thinking-NVFP4 — nvidia · ⬇️ 43,280 · ❤️ 33
  • Tags: Model Optimizer, safetensors, deepseek_v3, nvidia, ModelOpt
• EleutherAI/bergson-wikitext-2-gpt2 — EleutherAI · ⬇️ 5 · ❤️ 0
  • Tags: safetensors, influence-functions, interpretability, license:apache-2.0, region:us

📄 New model / evaluation papers (arXiv cs.CL + cs.AI)

No new model release papers found this week.

🆕 New benchmarks detected (not in benchmark_registry.json)

No new benchmarks detected beyond our current registry.

🏆 Papers With Code SOTA updates

No new SOTA entries this week.

✅ Suggested actions

• Review new HF models above — add any to results/benchmark.csv as new columns
• Check new arXiv papers for benchmark scores to add to frontier baselines
• Evaluate new benchmarks against your H100 pipeline compatibility
• Update docs/benchmark_registry.json if new benchmarks are confirmed valuable

---

Triggered by .github/workflows/monitor.yml · View workflow