Hackernews posts about SOTA

1. Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model (www.kimi.com)
   502 points by nekofneko 26 days ago | 239 comments
2. Ask HN: Open Models are 9 months behind SOTA, how far behind are Local Models?
   11 points by myk-e 13 days ago | 12 comments
3. Show HN: NSED is public – Mixture-of-Models to Hit SOTA using self-hosted AI (github.com)
   3 points by t_peersky 3 days ago | discuss
4. MiniMax's new open M2.5 and M2.5 Lightning near SOTA, costing 1/20th Opus 4.6 (venturebeat.com)
   3 points by gmays 5 days ago | discuss
5. Qwen3-coder-next: SOTA open source coding model (huggingface.co)
   3 points by binsquare 18 days ago | discuss
6. Ask HN: Do you think China will produce a SOTA model in the next 2 years
   2 points by upmind 4 days ago | 2 comments
7. Show HN：AgentHub - Unified, Stateful SDK for All SOTA LLMs and Agent Executions (github.com)
   2 points by PrismShadow 26 days ago | 1 comments
8. MiniMax M2.5 SOTA in Coding and Agent, Designed for Agent Universe (www.minimax.io)
   2 points by virgildotcodes 8 days ago | discuss
9. SotA ARC-AGI-2 Results with REPL Agents (www.symbolica.ai)
   2 points by tosh 10 days ago | discuss
10. Z.ai GLM-OCR: SOTA performance, optimized for complex document understanding (ocr.z.ai)
    2 points by pretext 19 days ago | discuss
11. CPU cloth simulation performance comparable to GPU SotA (sig25ddmpd.github.io)
    1 points by gsf_emergency_6 10 days ago | 2 comments
12. Ask HN: Views on Mastra's SOTA Memory?
    1 points by manthangupta109 9 days ago | 1 comments
13. Sanskrit AI beats CleanRL SOTA by 125% (huggingface.co)
    1 points by prabhatkr 14 days ago | 1 comments
14. Why SOTA Hybrid Models are Stuck at 1:7 (gist.github.com)
    1 points by eric2675 17 days ago | 1 comments
15. Realtime video model with SOTA quality (twitter.com)
    1 points by metarmeg 26 days ago | 1 comments
16. SoTA LLM Guardrails by Trusting the Typical [ICLR 2026] (arxiv.org)
    1 points by barthelomew 17 days ago | discuss
17. llm-d 0.4: Achieve SOTA performance across accelerators (llm-d.ai)
    1 points by teleforce 19 days ago | discuss
18. Show HN: SOTA NLP Models (huggingface.co)
    1 points by ChadNauseam 22 days ago | discuss
19. Kimi K2.5 – new open weights SOTA (huggingface.co)
    1 points by nikhizzle 26 days ago | discuss
20. Ask HN: What's the current best local/open speech-to-speech setup?
    265 points by dsrtslnd23 30 days ago | 71 comments
21. Show HN: Ourguide – OS wide task guidance system that shows you where to click (ourguide.ai)
    52 points by eshaangulati 26 days ago | 22 comments
22. Show HN: Honcho – Open-source memory infrastructure, powered by custom models (github.com)
    8 points by vvoruganti 25 days ago | discuss
23. Prediction: Claude 5 will be a major regression
    3 points by cadabrabra 19 days ago | 12 comments
24. Ask HN: A proposal for interviewing "AI-Augmented" Engineers
    3 points by vanbashan 19 days ago | 5 comments
25. Show HN: 17MB pronunciation scorer beats human experts at phoneme level
    3 points by fabiosuizu 1 day ago | 2 comments
26. Show HN: Free On-Device AI SDK to Run PyTorch on Mobile NPUs (Open Source) (github.com)
    3 points by kimys1324 24 days ago | discuss
27. Show HN: AgentHub – A unified SDK for LLM APIs with faithful validation (github.com)
    2 points by PrismShadow 28 days ago | discuss
28. Show HN: I could build the translation tool I've wanted for 10 years
    1 points by luxpir 25 days ago | 1 comments
29. Show HN: Custom ML model – if Spotify was instrumental (instr.io)
    1 points by day6 6 days ago | discuss
30. Show HN: Roe.md generate your own OpenClaw-like bot from a single Markdown file (github.com)
    1 points by guld 9 days ago | discuss