one cloud for compute, inference, and agents Question: https://nitter.cf/t.co/dOcscX4hmx Japan: @gmicloudJapan
Mountain View, CA
Joined December 2023
- Tweets3.7K
- Following62
- Followers10.2K
- Likes3.5K
Pinned Tweet
Introducing Agentbox v2:
Sandboxes for AI agents, on the same platform as GMI's models.
Your agent can run code, keep files, and pick up where it left off. Every sandbox is its own microVM, and every command returns stdout, stderr, and an exit code.
Early access opened last week. Today it's public.
Join us for a GMI Clouder x Factory Guild ambassador-led workshop 🛠️
📅 Nov 12, 9:30 AM PT
@kunal_twts and @adityawaslost, both GMI Clouders and Factory Guild members, walk through what they shipped with Droid on the GMI Cloud API.
Stay till the end for free GMI Cloud inference and @factory credits.
👉 luma.com/gmicloud-dbd1?utm_s…
GMI Cloud retweeted
The entire 3D model was built with Three.js r187, and the total cost was just $3.50.
OpenCode + DeepSeek v4.1 Flash built it via @gmi_cloud and it didn't use any special skills.
I'm genuinely amazed at what such an affordable model can accomplish.
🤖 Made with AI
Happy Halloween
If you are looking for a spooky game tonight, we made one for you: Dino Peekaboo
The first 10 ppl who finished the game, send your screenshot below, and we will give you some voucher credits as rewards!
enjoy: kentongmi.itch.io/dino-peeka…
GMI Cloud retweeted
2026/10/29台北101で開催されたGMI CLOUD DAY APAC2026はフィジカルAI業界の豪華ゲストがたくさん登場する凄いイベントでこれからGPUパワーで世界の経済が動く縮図を体験できるイベントでした。Andy様にもご挨拶できて良かったです。日本と台湾のGMI社員様のおかげで大変素晴らしい経験をさせていただきありがとうございます😊起業家向けのスケールという起業化支援サービスにも申し込んで今後GMI Cloud様のイベントにもたくさん参加していきます。ぜひよろしくお願いします。
@gmicloudJapan @gmi_cloud #GMI #Cloud #GPU
Join our creator tools hackathon and win 5K+ in cash prizes!
Hack the Feed, Fri Nov 6, 1–6pm at @Amplitude_HQ.
Build with GMI inference credits, @oriane_xyz’s video intelligence API, and @lovable_dev
Register here: luma.com/92iq2a12
GMI Cloud retweeted
One photo. 24 chat stickers.
I built meme with GPT Image 2.5 on @gmi_cloud:
- 17 styles
- English/Chinese captions
- PNG + ZIP downloads
Try it: meme.claptrap.chat
A GMI Clouder project.
#GMICloud #BuildInPublic
Our 2 takes on @ArtificialAnlys Intelligence Index v5:
1. 60% of the score now comes from private test data. Harder for models to train with the test data.
2. The median frontier model now uses 59k output tokens per task, from 35k.
that means infra needs predictable throughput and cost to keep up with longer, multi-step runs.
more 👇
Announcing Artificial Analysis Intelligence Index v5, featuring 2 new benchmarks and 3 upgrades. For the first time, the majority of the Index is built on private datasets
The Artificial Analysis Intelligence Index is our composite measure of model capabilities across tasks reflecting real-world usage today. v5 is our biggest upgrade since v4 launched in January. With v5, 60% of the Index is now based on evaluations with private test sets or held-out answers.
Alongside Intelligence Index v5, we’re launching the Artificial Analysis Progress Index v1, which tracks model capability over time on one continuous scale.
New to the Intelligence Index:
➤ FrontierCode 2.0: Cognition's newly-launched private benchmark of real-world coding tasks, graded on whether a repository maintainer would merge the agent's change
➤ Terminal-Bench-Science: 70 agentic scientific research tasks, run in a terminal environment
Upgraded:
➤ HLE-Diamond (replacing Humanity's Last Exam): a refined 1,000-question version of Humanity's Last Exam, split evenly between reasoning and knowledge questions
➤ AA-Omniscience v1.1: An updated dataset with saturated and public set questions removed, new questions added, and ambiguous or erroneous questions revised
➤ CritPt v2: a private, 40-question hard subset of CritPt v2, revised by the CritPt authors and run with a code-execution tool
Staying in the Index: AA-Briefcase, AutomationBench-AA, GDP.pdf and Terminal-Bench 4.0
Removed from the Index: GDPval-AA v2.1, SciCode, and AA-LCR v1.1
Key results:
➤ Claude Opus 5.5 (max) leads at 58, ahead of GPT-6 Astra (max, 57) and GPT-6.1 Sol (max, 55): Anthropic and OpenAI hold the top five models. Gemini 4 Argon (high) is next at 50
➤ The open weights gap widens from 3 points to 12: Kimi K3.1 Pro (weights not yet released) is the top open weights model at 46. Among released weights, DeepSeek V4.1 Flash 1017 (max) and GLM-5.3 (max) lead at 32 - 26 points behind the leader and below Claude Haiku 5.5 (max, 36)
➤ Cost per task varies by more than 13x among the leading models: of the 10 highest-scoring models shown, GPT-6.1 Terra (max) is the cheapest at $0.96 per Intelligence Index task and Claude Fable 5.1 (max) the most expensive at $13.04
