The Rio 3.5 model broke the internet this week. The plot twist? Itโs essentially our open-source model, Nex N2 Pro, wearing a different hat.
๐คฏ We analyzed the weights, and the recipe is exact: Rio 3.5 โ 0.6 * Nex N2 Pro + 0.4 * Qwen 3.5
It even literally introduces itself as "Nex N2 Pro" if you ask it without initial system prompt!
๐ We are flattered that the City of Rio used our work to achieve SOTA performance. Thanks for the ultimate benchmark validation.
๐ค But in the open-source world, attribution matters.
๐ Full mathematical proof & verify script in the first reply!
๐ข Nex-N2 is here!
A family of agentic models that doesn't just think, it acts!
Coding, search, tool use. All fused into a single agentic reasoning loop.
- Adaptive Thinking, auto-scales reasoning depth per step. Saves ~20% tokens, zero performance loss.
- Coherent Thinking, one thinking paradigm across search, coding, and tool use. No more fragile mode-switching.
๐ Result: Tier-1 open-source performance on SWE-bench, Terminal-Bench, GDPval, and more, tracking GPT-5.5 and Opus 4.7.
๐ Open-weight. Try it now.
๐ https://t.co/7oLSfyOCxB
๐ฆ https://t.co/c2CGhXWaz6
https://t.co/KJYXZIpk8M
https://t.co/vcjdZ9cuB6
๐ข Nex-N2 is here!
A family of agentic models that doesn't just think, it acts!
Coding, search, tool use. All fused into a single agentic reasoning loop.
- Adaptive Thinking, auto-scales reasoning depth per step. Saves ~20% tokens, zero performance loss.
- Coherent Thinking, one thinking paradigm across search, coding, and tool use. No more fragile mode-switching.
๐ Result: Tier-1 open-source performance on SWE-bench, Terminal-Bench, GDPval, and more, tracking GPT-5.5 and Opus 4.7.
๐ Open-weight. Try it now.
๐ https://t.co/7oLSfyOCxB
๐ฆ https://t.co/c2CGhXWaz6
https://t.co/KJYXZIpk8M
https://t.co/vcjdZ9cuB6
Introduce NexA4A: Automating Agent Construction with Natural Language.
Scaling the next generation of agentic environments:
https://t.co/4JfrZ8by0I
https://t.co/uI5PrZ0ECl
We rebuilt Grouped Per-Token Quantization at the kernel level โ removing the real bottleneck in FP8/MoE training and pushing efficiency to the hardware limit:
โข 20ร faster core ops
โข ~2.7 TB/s effective throughput
Details at: https://t.co/vLjipcS6JV
Meet Nex by NEX-AGI โ a non-thinking model built for agents that crushes it in coding, tool use, and roleplay ๐
โ SOTA among open models on Tau2-Bench, BFCL V4, GAIA2
โ Top-tier in frontend, vibe coding, and mini-program/backend dev (human eval confirmed)
โ Plug-and-play with Claude Code, Cursor, etc.
๐ Expolore Nex: https://t.co/8hhCIoFqRO
๐ฅ Free for now on SiliconFlow: https://t.co/juqcXci8bz
Welcome Nex-N1, a new series of agentic foundational models, to @huggingface
- available in different sizes from 8B, 30B, 32B to 671B
- strong in tool-use, web-search and real-world agentic workflow
- some SFT dataset has been open sourced
Technical report come up soon!