Building the future of work with AI
Upstage boosts work efficiency with industry-leading document processing engines and large-language models (LLMs).upstage.ai San Francisco, CAJoined September 2020
Thank you @thenewstack for the thoughtful deep dive on Solar Pro 4. Like they said — Consistent across turns, strong on long-context, ~90% cheaper per workflow. The workhorse for real work.
Upstage AI's Solar Pro 4 targets reliable AI agents for enterprise workflows, promising 90% lower costs than frontier models for document tasks. thenewstack.io/upstage-solar-…
🎉 We just hit #10 on the @OpenRouter LLM Leaderboard (Closed models, Today)
Solar Pro 4 is now sitting alongside the biggest names in the world — and that’s entirely thanks to you.
To everyone who’s been building with Solar Pro 4: thank you, sincerely. Every token you ran got us here. @NousResearch
We’re still in our promo period, so keep it coming — let’s climb even higher together. 🚀🌞
#SolarPro4#Upstage#OpenRouter
Solar Pro 4 is #2 on @OpenRouter's value leaderboard right now.
That board scores capability per dollar: @ArtificialAnlys benchmarks over the posted input/output rate. Only models in the top half of the benchmarks qualify at all — Solar Pro 4 lands in the top third on agentic (67th percentile), and the top 40% on intelligence and coding (63, 64). 524K context.
And it's #2 because ✨we're running 90% off✨ through Sept 10. $0.03 in, $0.12 out per 1M.
Which makes this the cheap month to find out whether it holds up on your workload.
Go break it 👉 openrouter.ai/upstage/solar-…#SolarPro4#Upstage#OpenRouter
Thanks @MervinPraison!
Thanks for the clear guide on using the Hermes agent! It'll be super helpful for newbies. As Mervin said, it's free until August 18, so hurry up and give it a try!
Solar Pro 4 is FREE until Aug 18 — and the best way to use it is Hermes Agent Desktop
Full beginners setup (step by step), 200k+ GitHub ⭐:
Telegram, Discord, Slack, WhatsApp built in
Kanban board — drag tasks, agent executes
Daily automations at 8am via Cron
Auto-creates its
🌠 Solar Pro 4 is now officially listed on Artificial Analysis. Intelligence Index: 42, up from Solar Pro 3's 14. A 3× gain in one generation. It finishes terminal tasks, keeps its footing across many tool calls, and stays accurate deep into long documents.
Try it free on Nous Research’s Hermes Agent through August 18. -> go.upstage.ai/4hLo3l1
Build with it at 90% off on Upstage Console and OpenRouter through September 10.
Thanks! @ArtificialAnlys
Korean AI lab Upstage has released Solar Pro 4, scoring 42 on the Artificial Analysis Intelligence Index, a significant increase from Solar Pro 3’s 14
Solar Pro 4 is @upstageai's new proprietary flagship reasoning model, replacing Solar Pro 3 from April 2026. At 42 on the
Free for one week, exclusively in Nous Portal:
Solar Pro 4, the new agentic model from @upstageai, built to complete multi-step tasks across long documents and to recognize when it needs more information rather than inventing it.
Try it today at portal.nousresearch.com
Solar Pro 4 is live.
Built for one job: production agents that finish the work.
▸ GDPval-AA (real work deliverables): 39
▸ τ³-Banking (policy retrieval + tool use): 23
▸ Terminal-Bench 2.1: 57
🪙 $0.30 / $1.20 per 1M tokens · $0.06 cached
90% off on Upstage Console and
Solar Pro 4 by @upstageai is the first Korean Lab to hit the Arena leaderboards for Agent Arena, Code Arena: WebDev, and Text Arena!
Solar Pro 4 debuts in the Agent Arena at #44 on par with models like: MiniMax M2.7 and Nemotron 3 Ultra.
Agent Arena uses causal tracing methodology to measure a model's net improvement across real-world, long-horizon agentic workflows. It indicates how much it improves outcomes relative to the average model.
Solar Pro 4 has an overall-net improvement score of -12.10%. By signal it landed:
- Confirmed Success: -7.3%
- Praise vs. Complaint: -23.4%
- Steerability: -13.6%
- Bash Recovery: -16.5%
- Tool Hallucination: +0.3%
Congrats to @upstageai for their contribution to the AI ecosystem!
Solar Pro 4 is live.
Built for one job: production agents that finish the work.
▸ GDPval-AA (real work deliverables): 39
▸ τ³-Banking (policy retrieval + tool use): 23
▸ Terminal-Bench 2.1: 57
🪙 $0.30 / $1.20 per 1M tokens · $0.06 cached
90% off on Upstage Console and
Solar Pro 4 is live.
Built for one job: production agents that finish the work.
▸ GDPval-AA (real work deliverables): 39
▸ τ³-Banking (policy retrieval + tool use): 23
▸ Terminal-Bench 2.1: 57
🪙 $0.30 / $1.20 per 1M tokens · $0.06 cached
90% off on Upstage Console and OpenRouter through Sep 10.
How it’s trained to finish jobs → go.upstage.ai/4cy3VPz
🎟️ 90% off on Upstage Console and @OpenRouter through Sep 10. $0.30 / $1.20 per 1M tokens · $0.06 cached How it’s trained to finish jobs → go.upstage.ai/4capBBe
Introducing #UpstageStudio by @upstageai
Processing documents is easy in 2026. A Claude Skill pulls fields from an invoice in a minute. Open-source parsers run locally.
So why build a document agent platform? Because the second document is a different problem.
Three walls everyone hits:
𝗔𝗰𝗰𝘂𝗿𝗮𝗰𝘆 — demos hold at 10 pages, collapse on 300.
𝗥𝗲𝗹𝗶𝗮𝗯𝗶𝗹𝗶𝘁𝘆 — 10 docs work, 10,000 crash the pipeline.
𝗖𝗼𝗺𝗽𝗹𝗶𝗮𝗻𝗰𝗲 — “the model said so” doesn’t fly in regulated industries.
We know these walls because we’ve spent five years walking into them.
Studio runs on OCR and LLMs processing 3M+ pages a day.
Every output traceable to its source. SaaS, or private in your AWS/Azure/on-prem with zero data exposure.
Bring us your second document: studio.upstage.ai#DocumentAgents#UpstageStudio
#SolarPro3 is now free to try. Stronger reasoning and instruction following — with the same API and a seamless upgrade from Solar Pro 2.
Free until Mar 2 🚀 Try it now: upstage.ai/blog/en/solar-…
139 Followers 5K FollowingFounder of Ganzo: Multi-Agents with local, remote, and distributed inference on a secure, anonymous, safe, private, uncensored Agentic Blockchain @ganzoai
267 Followers 2K Following🇬🇧🇵🇰 MLE @ DPD | Lead DS @ Daraz (Alibaba Group) |
Building NeighbourFit, WeaveSkip, SortMoments
Just a data point to be trained on, a vector of madness
488 Followers 1K FollowingLearning more than executing atm. All in in AI.
“You make your own luck if you stay at it long enough.”
Human-written tweets.
55 Followers 209 Following@Replit Ambassador
PhD in Engineering, Seoul National University
Private Instructor, Blue Ribbon EDU
Founder and Owner of @FiveStudy_Off
2K Followers 7K FollowingI like to build what I need.
Currently building Vespera.
Freedom-tech · Local-first · Korea
ROK-US-JP 🇰🇷🇺🇸🇯🇵
INVESTIGATE A-WEB
글은 하이라이트를 참고해주세요 🫶
3 Followers 37 FollowingNo open ports. Just connect. Secure HTTP, TCP & MCP tunnels for developers and AI teams. Reach local systems without opening inbound ports. By Qualiteg.
186 Followers 916 FollowingAI agent forums & social https://t.co/KLKIjgbFJn
Agent community-built projects include improving the English language (Ainglish), agent reproduction, AI museum.
153 Followers 271 FollowingA product of Fauci’s two experiments. Fauci said “the benefits of such experiments and the resulting knowledge outweigh the risks” now he takes the 5th amend.💉
6K Followers 1K FollowingThe world's largest gathering of insurance leaders and innovators.
🗓 September 29 - October 1, 2026
📍 Mandalay Bay, Las Vegas, NV
#ITCVegas
19K Followers 3K FollowingMachine learning & AI podcaster, community builder and all around enthusiast. Creator of the @TWIMLAI Podcast, TWIMLcon, TWIMLfest & the TWIML Solutions Guide.
2.6M Followers 47 FollowingThe official handle for NVIDIA. Blog: https://t.co/JAn5eKOTBT Support: https://t.co/6ln5FVnA2o All our social media: https://t.co/Uc56dL57Dh
5.1M Followers 4 FollowingOpenAI’s mission is to ensure that artificial general intelligence benefits all of humanity. We’re hiring: https://t.co/dJGr6LgzPA
596K Followers 569 FollowingFounder of the world’s most read daily AI newsletter @therundownai. Sharing the latest developments in the world of artificial intelligence.
1.8M Followers 1K FollowingCo-Founder of Coursera; Stanford CS adjunct faculty. Former head of Baidu AI Group/Google Brain. #ai #machinelearning, #deeplearning #MOOCs
74K Followers 7 FollowingThe best AI, all in one place. GPT-5.4, Grok 4.2, Claude 4.6, Veo 3 and more. At https://t.co/N5CdbrNcSa, or for iOS, Android, Mac, or Windows at https://t.co/YyFq8FS0ph
24K Followers 1K FollowingAccelerating humanity's transition to AGI — Note: we are exclusively located in Hillsborough, CA, no physical location in SF nor any other city yet.