Will Sun retweeted
๐๐ ๐ง๐๐ฌ ๐๐ข๐๐๐จ ๐.๐ ๐๐ง๐ญ๐๐ซ๐ฌ ๐ญ๐ก๐ ๐๐ซ๐ญ๐ข๐๐ข๐๐ข๐๐ฅ ๐๐ง๐๐ฅ๐ฒ๐ฌ๐ข๐ฌ ๐๐จ๐ฉ ๐๐
Agnes Video 2.5 is now ranked #17 on the Artificial Analysis Text to Video Leaderboard, with an Elo score of 1,082 and API pricing at $1.50/min.
Next up: Agnes Image 2.5 is also coming to the Artificial Analysis leaderboard.
artificialanalysis.ai/video/โฆ
Will Sun retweeted
๐๐ ๐ง๐๐ฌ ๐๐ ๐๐ฌ ๐๐จ๐ฐ ๐๐ฏ๐๐ข๐ฅ๐๐๐ฅ๐ ๐จ๐ง ๐๐ฉ๐๐ง๐๐จ๐๐
Connect Agnes AI directly to your coding workflow and start building, debugging, and iterating from OpenCode.
Another way to bring Agnes closer to where developers work.
Will Sun retweeted
๐๐๐ฆ๐ ๐๐ซ๐จ๐ฆ๐ฉ๐ญ. ๐๐ข๐๐๐๐ซ๐๐ง๐ญ ๐๐ ๐๐ฎ๐ข๐ฅ๐๐ฌ.
A simple idea: a retro typewriter experience.
Four AI models turned the same prompt into interactive demos, with different designs, animations, and details.
Which one would you choose?
Built with Agnes.
Will Sun retweeted
๐๐๐ฆ๐ ๐ฉ๐ซ๐จ๐ฆ๐ฉ๐ญ. ๐๐ข๐๐๐๐ซ๐๐ง๐ญ ๐๐ ๐๐ฎ๐ข๐ฅ๐๐ฌ.
We asked Agnes 2.5 Flash and DeepSeek V4 Flash to create the same interactive physics simulation from scratch.
See how different AI models approach coding, design, and real-time experiences.
Built with Agnes.
Will Sun retweeted
๐๐ ๐ง๐๐ฌ๐๐จ๐๐ ๐๐ฌ ๐๐จ๐ฐ ๐ข๐ง ๐๐ ๐๐จ๐๐
Bring AgnesCode directly into your editor.
Chat, edit, explain, and debug code without leaving VS Code, powered by Agnes 2.5 Flash.
Now available on the VS Code Marketplace. ๐
marketplace.visualstudio.comโฆ
Will Sun retweeted
๐ ๐๐ฌ๐๐ซ ๐๐ฎ๐ข๐ฅ๐ญ ๐ ๐๐ฎ๐๐ ๐๐ญ ๐๐ฉ๐ฉ ๐ฐ๐ข๐ญ๐ก ๐๐ ๐ง๐๐ฌ ๐.๐ ๐
๐ฅ๐๐ฌ๐ก
A user wanted a simpler way to manage monthly spending. Agnes 2.5 Flash helped turn that idea into a functional budget dashboard with expense tracking, subscriptions, spending limits, and smart insights.
From everyday problems to usable products.
Build with Agnes.
Will Sun retweeted
๐๐ด๐ป๐ฒ๐ ๐ฎ.๐ฑ ๐ฃ๐ฟ๐ผ ๐๐น๐ฝ๐ต๐ฎ ๐๐ฌ ๐๐จ๐ฐ ๐จ๐ง ๐ญ๐ก๐ ๐๐ซ๐ญ๐ข๐๐ข๐๐ข๐๐ฅ ๐๐ง๐๐ฅ๐ฒ๐ฌ๐ข๐ฌ ๐๐๐๐๐๐ซ๐๐จ๐๐ซ๐
๐๐ด๐ป๐ฒ๐ ๐ฎ.๐ฑ ๐ฃ๐ฟ๐ผ ๐๐น๐ฝ๐ต๐ฎ has been independently evaluated by Artificial Analysis, achieving an ๐๐ง๐ญ๐๐ฅ๐ฅ๐ข๐ ๐๐ง๐๐ ๐๐ง๐๐๐ฑ ๐ฌ๐๐จ๐ซ๐ ๐จ๐ ๐๐.
Coding emerged as a particular strength, with a ๐๐จ๐๐ข๐ง๐ ๐๐ง๐๐๐ฑ ๐ฌ๐๐จ๐ซ๐ ๐จ๐ ๐๐.๐. Combined with a ๐๐-๐ญ๐จ๐ค๐๐ง ๐๐จ๐ง๐ญ๐๐ฑ๐ญ ๐ฐ๐ข๐ง๐๐จ๐ฐ, full-modality input support, and accessible pricing, the results mark another important step forward for the models we train in-house.
It is also another meaningful milestone for a homegrown Singapore AI lab building globally competitive models.
Thank you to the Artificial Analysis team for the evaluation. Explore the full results in the original post.
Agnes AI has launched Agnes 2.5 Pro Alpha, a low-priced reasoning model reaching 39 on the Artificial Analysis Intelligence Index at $0.45/$0.90 per 1M tokens
Agnes AI (@agnesai_sapiens) is a Singapore-based AI lab that trains its own full-modality foundation models in-house across text, image, and video, and offers them through a free omni-modal API that has passed 3 million users. Agnes 2.5 Pro Alpha is a text, image, and video input reasoning model with text output.
Agnes 2.5 Pro Alpha scores 39 on the Artificial Analysis Intelligence Index, placing it mid-pack among reasoning models we have tested and below the current proprietary frontier. Its defining characteristic is price: at $0.45 per 1M input tokens and $0.90 per 1M output tokens, Agnes 2.5 Pro Alpha is among the lowest-priced proprietary models at its intelligence level.
Key results:
โค Agnes 2.5 Pro Alpha debuts at 39 on the Artificial Analysis Intelligence Index, its first appearance on our leaderboard. This places it mid-pack among current reasoning models.
โค Coding is a relative strength for Agnes 2.5 Pro Alpha, scoring 58.8 on the Coding Index, near the top for its intelligence tier. This is above similarly priced models including DeepSeek V4 Flash (56.2), GPT-5.4 mini (56.1) and Qwen3.7 Plus (55.9), and sits just behind Nex-N2-Pro (59.1).
โค On academic reasoning, Agnes 2.5 Pro Alpha is competitive for its tier, scoring 32% on Humanity's Last Exam and 88% on GPQA Diamond. Its Humanity's Last Exam result leads similarly priced proprietary models including GPT-5.4 mini (xhigh, 27%) and GPT-5.4 nano (xhigh, 26%), and its GPQA Diamond score is in line with peers near its Intelligence Index.
โค At $0.45/$0.90 per 1M input/output tokens, Agnes 2.5 Pro Alpha is the second cheapest model at its intelligence level. Blended at 7:2:1 (cache-input-output) it costs $0.18 per 1M tokens, behind only DeepSeek V4 Flash (Reasoning, Max Effort) at $0.06 among models scoring 38-40 on the Artificial Analysis Intelligence Index. Open weights alternatives including DeepSeek V4 Pro (Reasoning, Max Effort) and MiMo-V2.5-Pro match that $0.18 price at higher intelligence.
Additional model details:
โค Context window: 1M tokens.
โค Pricing: $0.45 / $0.90 / $0.0038 per 1M input / output / cache hit tokens.
โค Input modalities: Text, image, and video.
โค Availability: Agnes AI first-party API.
Will Sun retweeted
๐๐ฎ๐ข๐ฅ๐๐ข๐ง๐ ๐ญ๐ก๐ ๐
๐ฎ๐ญ๐ฎ๐ซ๐ ๐จ๐ ๐๐๐๐ฅ๐ญ๐ก๐๐๐ซ๐ ๐ฐ๐ข๐ญ๐ก ๐๐
At Hack4Health 2026, participants can use Microsoft Copilot Studio to access Agnes APIs and build solutions for real healthcare challenges.
Our Co-founder Alex will also join the event to share insights and connect with participants.
Will Sun retweeted
๐๐ง๐ ๐๐ซ๐จ๐ฆ๐ฉ๐ญ. ๐ ๐๐จ๐ฆ๐ฉ๐ฅ๐๐ญ๐ ๐๐ซ๐๐ฌ๐๐ง๐ญ๐๐ญ๐ข๐จ๐ง.
From a research paper to a polished, downloadable 8-slide PowerPoint, complete with methodology, charts, comparisons, conclusions, and future work.
Built with Agnes Code, powered by Agnes 2.5 Flash.
Will Sun retweeted
๐๐ด๐ป๐ฒ๐ ๐ฎ.๐ฑ ๐๐น๐ฎ๐๐ต ๐๐ฎ๐ซ๐ง๐ฌ ๐๐ง๐ ๐๐ซ๐จ๐ฆ๐ฉ๐ญ ๐๐ง๐ญ๐จ ๐ ๐๐จ๐ฎ๐ญ๐ข๐ช๐ฎ๐ ๐๐จ๐ญ๐๐ฅ ๐๐๐๐ฌ๐ข๐ญ๐.
Cinematic video, interactive room selection, hotel experiences, availability checks, pricing, and a complete reservation flow, all brought together in one polished website.
Build with Agnes.
Will Sun retweeted
๐๐ง๐ ๐๐ซ๐จ๐ฆ๐ฉ๐ญ. ๐
๐จ๐ฎ๐ซ ๐๐ ๐๐จ๐๐๐ฅ๐ฌ. ๐๐จ ๐๐๐ง๐ฎ๐๐ฅ ๐๐๐ข๐ญ๐ฌ.
We asked Agnes 2.5 Flash, ChatGPT 5.5 instant, Gemini 3.5 Flash, and Sonnet 5 to build the same interactive day-to-night scene.
Which one would you choose?
Build with Agnes.
Will Sun retweeted
๐๐ ๐ง๐๐ฌ ๐
๐ฎ๐ฅ๐ฅ-๐๐จ๐๐๐ฅ๐ข๐ญ๐ฒ ๐๐จ๐๐๐ฅ ๐๐๐ซ๐ข๐๐ฌ ๐๐จ๐ง๐ญ๐ข๐ง๐ฎ๐๐ฌ ๐ญ๐จ ๐๐ซ๐จ๐ฐ.
Weekly consumption reached ๐.๐๐๐ tokens, including ๐.๐๐๐ text tokens and ๐.๐๐๐ multimodal tokens.
Thank you to everyone building with Agnes AI.
Build with Agnes.
๐ค Made with AI
Will Sun retweeted
๐จ ้็ฃ
AMA ๆฅ่ขญ๏ผ
๐ฅ WAIC 2026 ่งๅฏไธๅไบซ๏ผๅ่ฏญ AI ็ๆ็นๆถๅป
โฐ ๆถ้ด๏ผ2026 ๅนด 7 ๆ 21 ๆฅ 21:00(UTC+8)
๐ ้พๆฅ๏ผteapaw.org/i/spaces/1lKQRRaโฆ
โจ ไธปๅๆน๏ผ@MetaEraCN๏ผ@AgentON_
๐๏ธ ไธปๆไบบ & ็น้ๅๅฎพ๏ผ@0xenid_๏ผ@JessicaMetaEra๏ผ@vito0526๏ผ@CyberYang00๏ผ@Deeplumen0922๏ผ@Genoptima๏ผ@Robinluo0920๏ผ@WillS725๏ผ@jackwaytoweb4
๐
่ฎพ็ฝฎ้น้๏ผไธๆไปฌไธ่ตทๆข็ดข๏ผ
- #WAIC ไธป่ฎบๅใๅ่ฎบๅ้ฝ่ฏดไบไปไน๏ผ
- WAIC ๆๅชไบๆๆๆ็ๅฑไฝ๏ผ
- ๆๅชไบๆฏ่พ็ซ็็่ต้๏ผ
- #WAIC2026 ๅๅ
ถไปๆตทๅ
ๅค AI ๆดปๅจ็ๅทฎๅผ
Will Sun retweeted
๐๐ด๐ป๐ฒ๐ ๐ฎ.๐ฑ ๐๐น๐ฎ๐๐ต ๐ญ๐ฎ๐ซ๐ง๐ฌ ๐จ๐ง๐ ๐ข๐๐๐ ๐ข๐ง๐ญ๐จ ๐ ๐๐จ๐ฆ๐ฉ๐ฅ๐๐ญ๐ ๐ฉ๐๐ญ ๐๐จ๐จ๐ ๐๐๐จ๐ฆ๐ฆ๐๐ซ๐๐ ๐ฐ๐๐๐ฌ๐ข๐ญ๐.
A single prompt generated responsive layouts, polished UI, product browsing, a shopping cart, and a complete checkout experience.
One Prompt Demo.
Build with Agnes.
Will Sun retweeted
๐๐ด๐ป๐ฒ๐ ๐ฎ.๐ฑ ๐๐น๐ฎ๐๐ต ๐ฏ๐ฌ ๐๐๐ ๐.๐-๐๐ง๐ฌ๐ญ๐๐ง๐ญ ๐ฏ๐ฌ ๐๐๐ฆ๐ข๐ง๐ข ๐.๐ ๐
๐ฅ๐๐ฌ๐ก ๐ฏ๐ฌ ๐๐ฅ๐๐ฎ๐๐ ๐๐จ๐ง๐ง๐๐ญ ๐.๐
One prompt. Four models. The exact same interactive Particle Gravity Playground challenge. No manual edits. No retries. Watch the results and decide for yourself.
Build with Agnes.
Will Sun retweeted
๐๐ ๐ง๐๐ฌ ๐.๐ ๐๐ซ๐จ, ๐.๐ ๐
๐ฅ๐๐ฌ๐ก & ๐.๐ ๐
๐ฅ๐๐ฌ๐ก ๐๐๐ซ๐จ๐ฌ๐ฌ ๐๐จ๐๐ข๐ง๐ ๐๐๐ง๐๐ก๐ฆ๐๐ซ๐ค๐ฌ
Our internal evaluation shows Agnes 2.5 Pro competing strongly across seven tests, while 2.5 Flash improves on 2.0 Flash in every benchmark, with the clearest gains on SWE Atlas.
Build with Agnes.
Will Sun retweeted
๐๐ ๐ง๐๐ฌ ๐.๐ ๐
๐ฅ๐๐ฌ๐ก ๐๐ฌ ๐๐จ๐ฐ ๐๐ข๐ฏ๐ ๐จ๐ง ๐๐ ๐ง๐๐ฌ ๐๐จ๐๐
Phased rollout begins today in ๐๐ ๐ง๐๐ฌ ๐๐จ๐๐โand itโs free to try. API access will roll out gradually throughout this week.
Generate, debug, refactor, analyze, and build complete products with fast reasoning and agentic workflows. More than code. Built to build anything.
agnes-ai.com/agnescode
๐ค Made with AI
Will Sun retweeted
๐๐ด๐ป๐ฒ๐ ๐ฎ.๐ฌ ๐๐น๐ฎ๐๐ต turns a language-learning idea into an interactive app.
Set goals, track progress, review vocabulary, and answer English quizzes with instant feedback.
One prompt becomes a learning flow people can use.
Build with Agnes.
Will Sun retweeted
๐๐ง๐ ๐๐ซ๐จ๐ฆ๐ฉ๐ญ. ๐๐ข๐ฑ ๐๐จ๐๐๐ฅ๐ฌ. ๐๐จ๐ฐ ๐๐จ๐ฆ๐ฉ๐ฅ๐๐ญ๐ ๐๐ฌ ๐ญ๐ก๐ ๐๐๐ฌ๐ฎ๐ฅ๐ญ?
Same coding prompt. No edits or follow-ups. ๐๐ด๐ป๐ฒ๐ ๐ฎ.๐ฌ ๐๐น๐ฎ๐๐ต delivered one of the most complete results.
๐๐ก๐ข๐๐ก ๐ซ๐๐ฌ๐ฎ๐ฅ๐ญ ๐๐จ ๐ฒ๐จ๐ฎ ๐ญ๐ก๐ข๐ง๐ค ๐ข๐ฌ ๐ญ๐ก๐ ๐๐๐ฌ๐ญ
Will Sun retweeted
๐๐ด๐ป๐ฒ๐ ๐ฎ.๐ญ ๐๐น๐ฎ๐๐ต ๐ฝ๐ฟ๐ฒ๐๐ถ๐ฒ๐ turns a single prompt into a complete interactive landing page for Luma Sleepbuds.
Product storytelling, color selection, soundscapes, sleep insights, routine building, and pre-order flows in one experience.
Build with Agnes.












