# ![@nexa_ai Avatar](https://lunarcrush.com/gi/w:26/cr:twitter::1684362265653567488.png) @nexa_ai NEXA AI

NEXA AI posts on X about ai, bounty, strong, inference the most. They currently have [-----] followers and [---] posts still getting attention that total [---] engagements in the last [--] hours.

### Engagements: [---] [#](/creator/twitter::1684362265653567488/interactions)
![Engagements Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1684362265653567488/c:line/m:interactions.svg)

- [--] Week [-----] -26%
- [--] Month [-------] -57%
- [--] Months [---------] +2,129%
- [--] Year [---------] +147%

### Mentions: [--] [#](/creator/twitter::1684362265653567488/posts_active)
![Mentions Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1684362265653567488/c:line/m:posts_active.svg)

- [--] Week [--] +16%
- [--] Month [--] +30%
- [--] Months [---] +863%
- [--] Year [---] +54%

### Followers: [-----] [#](/creator/twitter::1684362265653567488/followers)
![Followers Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1684362265653567488/c:line/m:followers.svg)

- [--] Week [-----] -0.32%
- [--] Month [-----] -7.10%
- [--] Months [-----] +236%
- [--] Year [-----] +263%

### CreatorRank: [---------] [#](/creator/twitter::1684362265653567488/influencer_rank)
![CreatorRank Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1684362265653567488/c:line/m:influencer_rank.svg)

### Social Influence

**Social category influence**
[stocks](/list/stocks)  56% [technology brands](/list/technology-brands)  56% [social networks](/list/social-networks)  3% [finance](/list/finance)  2%

**Social topic influence**
[ai](/topic/ai) 79%, [bounty](/topic/bounty) #487, [strong](/topic/strong) 11%, [inference](/topic/inference) 10%, [agentic](/topic/agentic) 8%, [the first](/topic/the-first) 8%, [apple](/topic/apple) 7%, [the official](/topic/the-official) 5%, [realtime](/topic/realtime) 5%, [discord](/topic/discord) 3%

**Top accounts mentioned or mentioned by**
[@qualcomm](/creator/undefined) [@snapdragon](/creator/undefined) [@nvidia](/creator/undefined) [@amd](/creator/undefined) [@ibm](/creator/undefined) [@hp](/creator/undefined) [@alanzhuly](/creator/undefined) [@android](/creator/undefined) [@ultralytics](/creator/undefined) [@mistralai](/creator/undefined) [@apple](/creator/undefined) [@nvidiaaipc](/creator/undefined) [@nexaai](/creator/undefined) [@liquidai](/creator/undefined) [@googleai](/creator/undefined) [@amdryzen](/creator/undefined) [@nvidias](/creator/undefined) [@zacklearner](/creator/undefined) [@alibabaqwen](/creator/undefined) [@alexchen4ai](/creator/undefined)

**Top assets mentioned**
[IBM (IBM)](/topic/ibm) [Alphabet Inc Class A (GOOGL)](/topic/$googl) [Microsoft Corp. (MSFT)](/topic/microsoft)
### Top Social Posts
Top posts by engagements in the last [--] hours

"🚀 This Week at Nexa: NexaML prefill optimization day-0 SOTA reasoning model support CES demo recap and On-Device Bounty Program. 1) NexaMLs latest prefill optimization on @Qualcomm Hexagon NPU delivers [--] faster prefill vs the onboard GPU [--] vs CPU and beats Apple M4 Max GPU by [---] on TTFT. With 0.6s TTFT at 1K tokens long-context RAG and agents finally feel instant. 2) The Nexa Qualcomm On-Device Bounty Program is heating up. We're inviting Android developers to build on-device AI apps on Qualcomm @Snapdragon with NexaSDK. Join by Feb [--] to win $6500 cash prizes Snapdragon devices plus"  
[X Link](https://x.com/nexa_ai/status/2014837795236544758)  2026-01-23T23:08Z [----] followers, [---] engagements


"We demoed Hyperlink Video Search in private beta at CES with @nvidia's latest RTX acceleration. The response was strong. Were now opening beta access to RTX [----] users. Search hours of footage like text and jump straight to the exact clip with precise timestamps instantly fully local powered by RTX acceleration on @NVIDIA_AI_PC. Join the waitlist to get access. (Link in thread) https://twitter.com/i/web/status/2015833750689743351 https://twitter.com/i/web/status/2015833750689743351"  
[X Link](https://x.com/nexa_ai/status/2015833750689743351)  2026-01-26T17:06Z [----] followers, [----] engagements


"@Qualcomm @Snapdragon 🙌🙌❤"  
[X Link](https://x.com/nexa_ai/status/2018819046419136891)  2026-02-03T22:49Z [----] followers, [--] engagements


"https://sdk.nexa.ai/model/YOLO26 https://sdk.nexa.ai/model/YOLO26"  
[X Link](https://x.com/nexa_ai/status/2021638985228202444)  2026-02-11T17:34Z [----] followers, [--] engagements


"You can now run GPT-OSS-20B locally on Qualcomm Hexagon NPU (partially CPU) using NexaSDK. Bring @OpenAIDevs's SOTA reasoning and agentic capabilities to your on-device mobile apps on @Snapdragon for the Nexa Qualcomm Bounty Program. Win $6500 cash prizes snapdragon device awards and partnership & marketing opportunities with @Qualcomm and Nexa AI. Join now and build the future of local agents 👇 https://twitter.com/i/web/status/2016198700767604950 https://twitter.com/i/web/status/2016198700767604950"  
[X Link](https://x.com/nexa_ai/status/2016198700767604950)  2026-01-27T17:16Z [----] followers, [---] engagements


"@Qualcomm 🙌🙌"  
[X Link](https://x.com/nexa_ai/status/2021670242733502891)  2026-02-11T19:38Z [----] followers, [--] engagements


"You can now use NexaSDK in Clawdbot(Moltbot) @openclaw your own personal assistant (seriously impressive work by @steipete) Clawdbot(Moltbot) is blowing up - AI can finally run and work for you 24/7. But with cloud AI costs add up fast. Local AI + open source fixes this. We set up Clawdbot(Moltbot) locally with NexaSDK on an @nvidia RTX [----] $0 token/inference fees plus personal data gets processed locally. 🦞Follow our cookbook to run your own Clawdbot(Molbot) locally for free forever (Link in thread) https://twitter.com/i/web/status/2016570341343240565"  
[X Link](https://x.com/anyuser/status/2016570341343240565)  2026-01-28T17:53Z [----] followers, [---] engagements


"The Nexa x @Qualcomm On Device Bounty Program is off to a great start In just [--] weeks we've had 50+ teams register to build the future of on-device AI. Participants are already building impressive Android projects pushing the frontier of what's possible with on-device AI powered by NexaSDK and @Snapdragon devices. From novel use cases to creative integrations this cohort is showing what happens when powerful models run locally. There's still time to join them. Submission deadline: February [--] Register now (Link in thread) https://twitter.com/i/web/status/2017723411263607196"  
[X Link](https://x.com/anyuser/status/2017723411263607196)  2026-01-31T22:15Z [----] followers, [---] engagements


"This Thursday come hang with us live. We're hosting a Live Q&A on the Qualcomm Developers Discord and YouTube Channel to break down the Nexa AI @Qualcomm On-Device Bounty Program. In [--] minutes we'll cover: Why on-device mobile AI matters right now Prizes timeline and how to join A live demo of what you can build with NexaSDK on Qualcomm @Snapdragon [--] Elite mobile devices Open Q&A bring your questions 📅 February [--] @ [--] AM PT Qualcomm Developers Discord & YouTube Whether you've already submitted a project or you're still exploring this is your chance to talk directly with the team building the"  
[X Link](https://x.com/anyuser/status/2018771705309937774)  2026-02-03T19:40Z [----] followers, [----] engagements


"https://www.youtube.com/watchv=7bSk4PsNl_0 https://www.youtube.com/watchv=7bSk4PsNl_0"  
[X Link](https://x.com/anyuser/status/2018771707675419073)  2026-02-03T19:40Z [----] followers, [---] engagements


"We're going LIVE on the Qualcomm Developers channel in just a few hours Join Nexa and @Qualcomm's on-device AI experts for a live Q&A on: The latest trends in local AI for mobile apps Next-gen on-device AI capabilities The Nexa x Qualcomm Builder Bounty Program and why now is the time to jump in Bring your questions. See you at [--] AM PT Feb 5👇 https://twitter.com/i/web/status/2019254185603264795 https://twitter.com/i/web/status/2019254185603264795"  
[X Link](https://x.com/anyuser/status/2019254185603264795)  2026-02-05T03:38Z [----] followers, [---] engagements


"https://www.youtube.com/watchv=7bSk4PsNl_0 https://www.youtube.com/watchv=7bSk4PsNl_0"  
[X Link](https://x.com/anyuser/status/2019254187587219766)  2026-02-05T03:38Z [----] followers, [---] engagements


"Catch @alanzhuly at @langflow_ai's Hacking Agents AI Meetup at @IBM San Francisco office at 5pm Feb [--]. Come connect and check out how Nexa AI is advancing local agents and MCP on NPU. Excited to be speaking at @langflow_ai's Hacking Agents AI Meetup this month at @IBM SF I'll be showcasing what we're building at @nexa_ai AI that runs locally on NPU pushing the boundaries of local agents. The lineup is stacked: OpenRAG MCP Context Forge IBM Bob and a Excited to be speaking at @langflow_ai's Hacking Agents AI Meetup this month at @IBM SF I'll be showcasing what we're building at @nexa_ai AI"  
[X Link](https://x.com/anyuser/status/2019556258563682568)  2026-02-05T23:38Z [----] followers, [---] engagements


"Incredible live Q&A for the Nexa x Qualcomm On-Device Bounty Program with @Qualcomm's @manojkhilnani and Derrick Johnson on the official Developers channel yesterday. @zacklearner and @alanzhuly nerded out on on-device mobile AI trends and capabilities ins and outs of bounty programs and walked through running RAG on mobile with Nexa SDK. The NPU vs CPU numbers speak for themselves. Catch the replay and can't wait to see how folks adapt it for creative use cases that revolutionize personal intelligence Submission is open till Feb 15th 👇 https://twitter.com/i/web/status/2019822532711559480"  
[X Link](https://x.com/anyuser/status/2019822532711559480)  2026-02-06T17:16Z [----] followers, [----] engagements


"https://sdk.nexa.ai/bounty https://sdk.nexa.ai/bounty"  
[X Link](https://x.com/anyuser/status/2019822534406164694)  2026-02-06T17:16Z [----] followers, [---] engagements


"The Nexa @Qualcomm On-Device Bounty Program is in full swing. To help teams move faster we open-sourced a local Therapy Notes sample app built with NexaSDK + @kotlin like Plaud AI but running fully on @Snapdragon-powered mobile. It turns 30-min therapy audio into structured SOAP notes using @nvidia Parakeet for ASR and @liquidai LFM2.5-1.2B + @Alibaba_Qwen Qwen3-4B for summarization all on device on Hexagon NPU with no privacy risk and no compliance overhead. Perfect example of where local mobile AI really shines. The tutorial covers: 0:000:26 Project overview 0:273:41 Running ASR + LLM"  
[X Link](https://x.com/nexa_ai/status/2020913294928916814)  2026-02-09T17:30Z [----] followers, [---] engagements


"https://github.com/vinovo/clinical-transcription-demo https://github.com/vinovo/clinical-transcription-demo"  
[X Link](https://x.com/anyuser/status/2020913298040860878)  2026-02-09T17:30Z [----] followers, [---] engagements


"6 days until the Nexa @Qualcomm On-Device Bounty Program Phase [--] deadline: Feb [--] 11:59 PM PT. Blocked or unsure what to submit Join our office hours this Friday Feb [--] [---] PM PT on Nexa AI Discord. Well help you troubleshoot debug and get across the line. Grab a 20-min slot or drop in anytime https://twitter.com/i/web/status/2021270439818035582 https://twitter.com/i/web/status/2021270439818035582"  
[X Link](https://x.com/nexa_ai/status/2021270439818035582)  2026-02-10T17:10Z [----] followers, [---] engagements


"https://calendar.app.google/XMwZXLvwS7DRNMWE6 https://calendar.app.google/XMwZXLvwS7DRNMWE6"  
[X Link](https://x.com/anyuser/status/2021270442359820453)  2026-02-10T17:10Z [----] followers, [--] engagements


"Tons of Nexa x Qualcomm On-Device Bounty Program participants pitched camera-native local AI ideas so NexaSDK now runs the full @ultralytics YOLO26 family on @Snapdragon [--] Elite. Accelerated by @Qualcomm Hexagon NPU this unlocks real-time low-power and private visual understanding that recognizes objects tracks human poses and detects rotated items. Ready to compete Build with YOLO26 for the On-Device Bounty Program and submit by Feb [--]. 👇 Expert support standing by https://twitter.com/i/web/status/2021638983521095794 https://twitter.com/i/web/status/2021638983521095794"  
[X Link](https://x.com/anyuser/status/2021638983521095794)  2026-02-11T17:34Z [----] followers, [---] engagements


"@ultralytics @Snapdragon @Qualcomm https://huggingface.co/NexaAI/yolo26s-npu-mobile https://huggingface.co/NexaAI/yolo26m-npu-mobile https://huggingface.co/NexaAI/yolo26l-npu-mobile https://huggingface.co/NexaAI/yolo26x-npu-mobile https://huggingface.co/NexaAI/yolo26s-npu-mobile https://huggingface.co/NexaAI/yolo26m-npu-mobile https://huggingface.co/NexaAI/yolo26l-npu-mobile https://huggingface.co/NexaAI/yolo26x-npu-mobile"  
[X Link](https://x.com/anyuser/status/2022013612387393912)  2026-02-12T18:23Z [----] followers, [--] engagements


"The Nexa x @Qualcomm On-Device Bounty Program is heating up 🔥 Phase [--] closes in [--] DAYS (Feb 15th 11:59pm PT) 90+ teams registered and we are seeing incredible ideas: Offline translator Multimodal assistants Productivity agents Accessibility tools We want to help you win and drive real impact on how people interact with AI. Every registered team gets 1-on-1 support from us and we're featuring all quality submissions to make sure your work gets seen. Don't sit on the sidelines. Build something real with NPU-powered on-device AI and win big rewards: $6500 in cash prizes + official partnership"  
[X Link](https://x.com/nexa_ai/status/2022355537166565506)  2026-02-13T17:01Z [----] followers, [---] engagements


"https://sdk.nexa.ai/bounty https://sdk.nexa.ai/bounty"  
[X Link](https://x.com/nexa_ai/status/2022355539414651246)  2026-02-13T17:01Z [----] followers, [--] engagements


"@TheAIColony @Qualcomm Intelligence comes from timing and awareness. When the system stays active and understands context in real time it stops being a voice interface and starts feeling like something thats actually helpful in the cabin"  
[X Link](https://x.com/nexa_ai/status/1996747593960108159)  2025-12-05T01:04Z [----] followers, [--] engagements


"@BJ_CogniVision @Qualcomm Real-time context understanding delivers ambient smart experience for users"  
[X Link](https://x.com/nexa_ai/status/1996771435105403389)  2025-12-05T02:39Z [----] followers, [--] engagements


"Thank you @AIatAMD for the feature. Seamless @PyTorch + ROCm compatibility isnt just a tooling win - it lowers friction for researchers and accelerates deployment paths for builders. Openness portability and real performance gains are what move the ecosystem forward. Were excited to keep pushing the stack together. More seamless compatibility between @PyTorch and ROCm means researchers can focus on modeling and kernel engineers can keep pushing performance. @alexchen4ai CEO of @nexa_ai shares why this matters for both sides of the stack. https://t.co/35TILdKq04 More seamless compatibility"  
[X Link](https://x.com/nexa_ai/status/1997010803204280728)  2025-12-05T18:30Z [----] followers, [---] engagements


"@Qualcomm AutoNeural on Qualcomm SA8295P NPU = responsive private and reliable next-gen in-car assistant. 🔥🤌"  
[X Link](https://x.com/nexa_ai/status/1997099996844527696)  2025-12-06T00:25Z [----] followers, [--] engagements


"@Qualcomm @GeelyAutoGroup @GeelyAutoGlobal Thank you for the great collaboration. Were excited to see how people push it in real production smart-cockpit experiences"  
[X Link](https://x.com/nexa_ai/status/1997104626005623275)  2025-12-06T00:43Z [----] followers, [---] engagements


"This week at Nexa AI: Nexa SDK for Linux launch AutoNeural-VL launch for next-gen smart cockpit Featured by Qualcomm blog and AMD talk new model support⚡ 1) @alexchen4ai CEO and founder of Nexa AI was featured by @AMD highlighting the impact of tighter PyTorch + ROCm integration for researchers and kernel engineers. 2) We launched NexaSDK for Linux with @Docker the first Linux SDK that unifies NPU + GPU + CPU with Day-0 support for top LLMs VLMs Embeddings Rerank ASR and CV models from @IBM @nvidia @ultralytics and @aiatmeta and @JinaAI_ powering next-gen IoT AI apps. 3) Meet"  
[X Link](https://x.com/nexa_ai/status/1997730573876646354)  2025-12-07T18:10Z [----] followers, [----] engagements


"AutoNeural resonated more than we anticipated. Since launch we crossed 870k views 3k likes across platforms and saw strong interest from industry leaders engineering teams and automotive OEMs. The automotive industry has been waiting for production-grade multimodal AI in the cabin. AutoNeural redefines what AI can do in the car with [--] lower latency (100ms) [--] higher visual detail (768) and [--] better accuracy running fully on @Qualcomm SA8295P NPU. We built AutoNeural to see understand and act on cabin and environmental context instantly and privately an ambient intelligence system that"  
[X Link](https://x.com/nexa_ai/status/1998075072972480665)  2025-12-08T16:59Z [----] followers, [---] engagements


"https://nexa.ai/book-a-call https://nexa.ai/book-a-call"  
[X Link](https://x.com/nexa_ai/status/1998075076567269570)  2025-12-08T16:59Z [----] followers, [---] engagements


"Ministral-3-3B just got 50% faster on @Qualcomm Hexagon NPU with the NexaML Turbo Engine were now hitting 27+ tok/s fully on-device. NexaML is the only inference engine running this latest @MistralAI model natively on NPU. That means state-of-the-art small-model intelligence with NPU-class latency and power efficiency ready for local copilots and offline apps across compute mobile XR automotive and IoT. Try today using NexaSDK"  
[X Link](https://x.com/nexa_ai/status/1998441874412155333)  2025-12-09T17:17Z [----] followers, [---] engagements


"👉 https://sdk.nexa.ai/model/%20Ministral-3-3B https://sdk.nexa.ai/model/%20Ministral-3-3B"  
[X Link](https://x.com/nexa_ai/status/1998441876014371284)  2025-12-09T17:17Z [----] followers, [---] engagements


"Self-driving is taking the work out of steering. But inside the cabin everything is still manual: juggling messages maps and apps while trying to stay focused on the road. We think in-car AI should feel like autopilot for the cabin.The car should understand whats happening on your screens and in your apps infer what you need and quietly take care of itwhile you stay in control. Thats what you can build with AutoNeural. In the example below a new message pops up: Party moved to Sunset View Winery." AutoNeural reads the card on the display understands it as a destination change finds the winery"  
[X Link](https://x.com/nexa_ai/status/1998828334638313559)  2025-12-10T18:53Z [----] followers, [----] engagements


"Introducing NexaSDK for iOS and MacOS run and build with the latest multimodal AI models fully on-device on @Apple Neural Engine GPU and CPU. This is the first and only SDK that enables developers to run the latest SOTA models on NPU across iPhones and Mac Laptops achieving [--] energy efficiency and [--] faster performance. With just [--] lines of code you can run state-of-the-art models across tasks including: Embedding: Nexa EmbedNeural for image search LLM: Granite-Micro (@IBM) Ministral3-3B (@MistralAI) Gemma3 (@GoogleAI) Qwen3-0.6B/4B (@Qwen) CV: PaddleOCR (@Baidu) ASR: Parakeet v3 (@NVIDIAAI)"  
[X Link](https://x.com/nexa_ai/status/1999163909765468243)  2025-12-11T17:06Z [----] followers, [--] engagements


"Introducing NexaSDK for iOS and MacOS run and build with the latest multimodal AI models fully on-device on @Apple Neural Engine GPU and CPU. This is the first and only SDK that enables developers to run the latest SOTA models on NPU across iPhones and Mac Laptops achieving [--] energy efficiency and [--] faster performance. With just [--] lines of code you can run state-of-the-art models across tasks including: Embedding: EmbedNeural for image search LLM: Granite-Micro (@IBM) Ministral3-3B (@MistralAI) Gemma3 (@GoogleAI) Qwen3-0.6B/4B (@Qwen) CV: PaddleOCR (@Baidu_Inc) ASR: Parakeet v3 (@NVIDIAAI)"  
[X Link](https://x.com/nexa_ai/status/1999170859085472160)  2025-12-11T17:34Z [----] followers, 12.3K engagements


"GitHub: Docs: https://docs.nexa.ai/nexa-sdk-ios/overview https://github.com/NexaAI/nexasdk-mobile-iOS-framework/tree/main https://docs.nexa.ai/nexa-sdk-ios/overview https://github.com/NexaAI/nexasdk-mobile-iOS-framework/tree/main"  
[X Link](https://x.com/nexa_ai/status/1999170861904138596)  2025-12-11T17:34Z [----] followers, [---] engagements


"Most driving stress doesnt come from steering. It comes from everything else in and around the car kids and pets street signs missing items everyday chaos. Todays self-driving copilots cant help with that. You need another pair of eyes that watches the cabin and the surrounding environment keeps track of whats happening and brings the right information to the driver at the right moment. Thats what you can build with AutoNeural. In the examples below - Parents dont need to constantly turn around AutoNeural alerts when kids in the back need attention - No more stress over missing items"  
[X Link](https://x.com/nexa_ai/status/1999525005827449087)  2025-12-12T17:01Z [----] followers, 26.1K engagements


"Contact us to explore a pilot: https://nexa.ai/book-a-call https://nexa.ai/book-a-call"  
[X Link](https://x.com/nexa_ai/status/1999525007563821211)  2025-12-12T17:01Z [----] followers, [---] engagements


"This week at Nexa AI AutoNeural momentum and use cases Apple ANE support on iOS & macOS and new model support 1) AutoNeural-VL-1.5B the first real-time multimodal model built for in-car AI crossed 880k+ views and 3k+ likes at launch with strong interest from automotive OEMs and engineering teams. It runs fully on the @Qualcomm SA8295P NPU with 100ms latency [--] finer visual detail (768) and up to [--] higher accuracy delivering production-grade in-cabin external and UI understanding for always-on smart cockpits. 2) NexaSDK is the first and only runtime that runs the latest multimodal AI models on"  
[X Link](https://x.com/nexa_ai/status/2000251153033076955)  2025-12-14T17:06Z [----] followers, [----] engagements


"The next generation of mobile apps will run multimodal AI locally by default. Today were making it practical for developers to ship. We just launched NexaSDK for Mobile on Product Hunt. Developers can run the latest multimodal AI models fully on-device in iOS & @Android apps with @Apple Neural Engine and @Qualcomm Hexagon NPU acceleration. In just [--] lines of code build chat multimodal search and audio features with no cloud cost complete privacy 2x faster speed and [--] better energy efficiency. Try out NexaSDK for Mobile and wed love your feedback (Product Hunt link in thread)."  
[X Link](https://x.com/nexa_ai/status/2000974949432107430)  2025-12-16T17:02Z [----] followers, 321.8K engagements


"Support our Product Hunt Launch: https://www.producthunt.com/products/nexasdk-for-mobile https://www.producthunt.com/products/nexasdk-for-mobile"  
[X Link](https://x.com/nexa_ai/status/2000974951076257832)  2025-12-16T17:02Z [----] followers, [---] engagements


"Hyperlink for @Windows is now live on the @MicrosoftStore. Your computer already has all your context its time to make it actually smart. Hyperlink is a private on-device AI agent that can understand whats in your computer files (docs notes PDFs screenshots) answer in natural language and cite exactly where each answer comes from. When the AI agent runs fully on-device accelerated by NPU/GPU/CPU you get speed + privacy + battery life + real personalization without uploading your info to the cloud. Agentic OS is the future: Hyperlink was featured and presented on stage at @Microsoft Ignite"  
[X Link](https://x.com/nexa_ai/status/2001353790050738302)  2025-12-17T18:08Z [----] followers, 44.4K engagements


"@JocelynRog29916 @Windows @MicrosoftStore Thanks for the support Let us know your feedback"  
[X Link](https://x.com/nexa_ai/status/2001405457475473883)  2025-12-17T21:33Z [----] followers, [--] engagements


"NexaSDK for Mobile got No.1 Product of the Day on @ProductHunt It received strong community support and was featured by the @Qualcomm Developer Blog as a simple way to bridge on-device AI to smartphones. Mobile AI is clearly shifting on-device but until now shipping multimodal AI on phones has been slow fragmented and battery-draining. NexaSDK for Mobile fixed this by letting developers run text vision and audio models fully on-device across iOS and @Android in just three lines of code. With native NPU acceleration it delivers [--] faster inference and [--] better energy efficiency than on CPU. Try"  
[X Link](https://x.com/nexa_ai/status/2001700005586080165)  2025-12-18T17:04Z [----] followers, 30.4K engagements


"On-device AI agents just got a real upgrade: @Google released FunctionGemma-270M tuned for function calling with unified action and chat and the model is built for customization. FunctionGemma is already supported in NexaSDK and runs blazingly fast. Run with one line of code: nexa infer NexaAI/functiongemma-270m-it-GGUF Download the latest NexaSDK and try it today👇 https://twitter.com/i/web/status/2002062655016825164 https://twitter.com/i/web/status/2002062655016825164"  
[X Link](https://x.com/nexa_ai/status/2002062655016825164)  2025-12-19T17:05Z [----] followers, [----] engagements


"NexaSDK is #1 on GitHub Trending today. 🏆 Its a sign developers want a unified inference engine that runs new multimodal models Day-0 on-device optimized across NPU/GPU/CPU. If you are an AI developer and haven't tried NexaSDK don't miss out: https://github.com/NexaAI/nexa-sdk https://github.com/NexaAI/nexa-sdk"  
[X Link](https://x.com/nexa_ai/status/2002635493742424406)  2025-12-21T07:01Z [----] followers, [---] engagements


"This week at Nexa AI NexaSDK No.1 trending on GitHub NexaSDK for Mobile ranked No.1 Product of the Day on Product Hunt Hyperlink on the Microsoft Store and the latest model support. 1) NexaSDK hit the No.1 trending repository on @GitHub indicating strong demand for a unified inference engine with Day-0 top-tier multimodal model support across NPU / GPU / CPU. 2) NexaSDK for Mobile launched on @ProductHunt as the No.1 Product of the Day received strong community support and was featured by the @Qualcomm Developer Blog as a simple way to bridge on-device AI to smartphones. It lets developers"  
[X Link](https://x.com/nexa_ai/status/2002789777742164110)  2025-12-21T17:14Z [----] followers, [---] engagements


"Celebrating Nexa SDK hitting 7K on @github. In the past month weve seen rapid growth and strong community recognition: Nexa SDK ranked No.1 on GitHub Trending and No.1 Product of the Day on @ProductHunt. That momentum signals clear demands for a unified inference engine that runs the latest top-tier multimodal models across NPU/GPU/CPU from laptops to mobiles automotive IoT and more. The shift is happening fast. More updates are coming very soon. https://twitter.com/i/web/status/2003150714814759308 https://twitter.com/i/web/status/2003150714814759308"  
[X Link](https://x.com/nexa_ai/status/2003150714814759308)  2025-12-22T17:08Z [----] followers, [---] engagements


"https://github.com/NexaAI/nexa-sdk https://github.com/NexaAI/nexa-sdk"  
[X Link](https://x.com/nexa_ai/status/2003150718228922534)  2025-12-22T17:08Z [----] followers, [---] engagements


"Nexa is showcasing Hyperlink with @HP at CES [----]. Hyperlink is an on-device AI super assistant and were demoing new agentic capabilities designed to change how people get work done. Under the hood it runs fully on the @Qualcomm Hexagon NPU on @Snapdragon powered by our NexaSDK inference engine for industry-leading speed and efficiency. https://twitter.com/i/web/status/2003510829539557684 https://twitter.com/i/web/status/2003510829539557684"  
[X Link](https://x.com/nexa_ai/status/2003510829539557684)  2025-12-23T16:59Z [----] followers, [---] engagements


"Nexa AI is coming back to #CES2026. In [----] we saw a clear shift: AI is moving on-devicedriven by privacy speed cost and real personalization. We shipped on that vision partnered across the ecosystem and grew a strong developer and industry community around it. Exciting partner announcements for CES are coming soon. Book a call with us and explore how ondevice extends whats possible with AI together: https://nexa.ai/book-a-call https://nexa.ai/book-a-call"  
[X Link](https://x.com/nexa_ai/status/2003875279618691195)  2025-12-24T17:07Z [----] followers, [---] engagements


"Merry Christmas and Happy Holidays from Nexa AI. [----] was a breakthrough year for on-device AI. Were excited to keep pushing the boundaries in [----]. Thank you for the support. More to come"  
[X Link](https://x.com/nexa_ai/status/2004243421226316053)  2025-12-25T17:30Z [----] followers, [--] engagements


"Merry Christmas and Happy Holidays from Nexa AI. [----] was a breakthrough year for on-device AI. Were excited to keep pushing the boundaries in [----]. Thank you for the support. More to come. https://twitter.com/i/web/status/2004245984558338408 https://twitter.com/i/web/status/2004245984558338408"  
[X Link](https://x.com/nexa_ai/status/2004245984558338408)  2025-12-25T17:40Z [----] followers, [---] engagements


"Big milestone for Nexa AI: well be showcased at @AMDs official CES booth. Hyperlink is also featured on AMDs official website as the industry-leading local AI super assistantrunning on our unified inference engine NexaSDK accelerated by @AMDRyzen AI. This is what agentic AI on your laptop is supposed to feel like. Learn more at: https://www.amd.com/en/ecosystem/isv/consumer-partners/nexa.html https://www.amd.com/en/ecosystem/isv/consumer-partners/nexa.html"  
[X Link](https://x.com/nexa_ai/status/2005685710355923429)  2025-12-29T17:01Z [----] followers, 23.2K engagements


"Were bringing Hyperlink to @nvidia's official CES booth. Your on-device AI assistant already understands your files. It now takes a major step forward with new multimodal agentic capabilities. This changes how your personal AI works. Dont miss it"  
[X Link](https://x.com/nexa_ai/status/2006049212312498321)  2025-12-30T17:06Z [----] followers, [---] engagements


"2025 marked a defining year for on-device AI and for Nexa AI. The shift to on-device is clear with Nexa AI leading the charge. In under [--] months we achieved: 7K+ GitHub stars No.1 GitHub Trending No.1 Product Hunt 40+ latest models brought to previously unsupported backends. Day-0 NPU support for IBM Granite Qwen3-VL Google EmbeddingGemma Mistral [--] and more. 20+ Partner announcements and featured at [--] major events including CES Microsoft Ignite Snapdragon Summit and more Driven by: NexaSDK + NexaML unified inference engine that revolutionizes local AI inference for the edge Hyperlink the"  
[X Link](https://x.com/nexa_ai/status/2006411787483418680)  2025-12-31T17:07Z [----] followers, [---] engagements


"A new year a new default for AI. Local. Private. On-device"  
[X Link](https://x.com/nexa_ai/status/2006779947596296281)  2026-01-01T17:29Z [----] followers, [---] engagements


"Mobile AI should be fast efficient and always available. Your phones NPU is built for that but most AI apps barely use it. Nexa Studio is the first local AI mobile app that runs SOTA multimodal AI (LLMs audio vision) fully on-device on @Qualcomm Hexagon NPUs optimized for speed battery life multimodal tasks and always-on use. In the demo: - Chat: @IBM Granite-4-Micro - Image captioning: OmniNeural-4B - Audio transcription: @nvidia's Parakeet-tdt-0.6b-v3 - Image search: EmbedNeural 100% local private offline. Accelerated on NPU. Link in thread."  
[X Link](https://x.com/nexa_ai/status/2007146735395844267)  2026-01-02T17:47Z [----] followers, [---] engagements


"Try it here: https://play.google.com/store/apps/detailsid=com.nexa.studio&pcampaignid=web_share https://play.google.com/store/apps/detailsid=com.nexa.studio&pcampaignid=web_share"  
[X Link](https://x.com/nexa_ai/status/2007146736901566772)  2026-01-02T17:47Z [----] followers, [---] engagements


"Were excited to join the official @Qualcomm booth at CES [----] to demonstrate faster cooler multimodal intelligence on-device on Qualcomm Hexagon NPUs. See demos from @memories_ai and @pyannoteAI powered by Nexa SDK Hyperlink's new multimodal capabilities and Nexa Studio the first local AI app that runs SOTA multimodal AI (LLMs audio vision) fully local on NPUs. Follow us to stay updated. https://twitter.com/i/web/status/2007497315100291394 https://twitter.com/i/web/status/2007497315100291394"  
[X Link](https://x.com/nexa_ai/status/2007497315100291394)  2026-01-03T17:00Z [----] followers, [---] engagements


"A snapshot of Nexa AIs presence at CES 2026: Powering the next generation of on-device multimodal AI. Nexa AI is powering a growing ecosystem of partners to demonstrate what multimodal AI looks like when it truly runs on device fast efficient and built to scale across hardware models and applications. Featured at: @Qualcomm booth: demos with @memories_ai and @pyannoteAI powered by Nexa SDK plus Hyperlink and Nexa Studio @nvidia booth: Hyperlink's new multimodal agentic capabilities accelerated by @NVIDIAGeForce @AMD booth: Hyperlink accelerated by @AMDRyzen @HP booth: Hyperlink on HP devices"  
[X Link](https://x.com/nexa_ai/status/2007862762299445686)  2026-01-04T17:12Z [----] followers, [---] engagements


"Were bringing Hyperlinks new multimodal agentic workflows to the official @Qualcomm booth at @CES. NPU-accelerated local inference + MCP lets an on-device agent do vision understanding + function calling across devices. Demo: capture an image on mobile send it to Hyperlink on your PC it understands what it sees use the correct MCP server to complete the action. The next era of cross-device personal intelligence starts here. Reach out or comment and we will share with you a demo. https://twitter.com/i/web/status/2008226244614299967 https://twitter.com/i/web/status/2008226244614299967"  
[X Link](https://x.com/nexa_ai/status/2008226244614299967)  2026-01-05T17:17Z [----] followers, [---] engagements


"NexaSDK now ships Day-0 support for @liquidais latest LFM2.5 on @Qualcomm Hexagon NPU Adreno GPU and Oryon CPU. One unified inference engine that runs across Compute Mobile IoT and Automotive on Android Microsoft Windows and Linux. Benchmarks: - @Snapdragon X Elite NPU (Compute): Prefill speed: [------] tok/s Decode speed: [----] tok/s - Snapdragon [--] Gen [--] NPU (Mobile): Prefill speed: [------] tok/s Decode speed: [----] tok/s - Dragonwing IQ-9075 NPU (IoT): Prefill speed: [------] tok/s Decode speed: [----] tok/s Why LFM2.5 matters: Best quality at 1B size while staying memory-efficient and fast and"  
[X Link](https://x.com/nexa_ai/status/2008387626383143359)  2026-01-06T03:58Z [----] followers, 25.7K engagements


"@liquidai Congrats on the launch and excited to parter with you bring day-0 support on Qualcomm NPU in NexaSDK: https://github.com/NexaAI/nexa-sdk https://github.com/NexaAI/nexa-sdk"  
[X Link](https://x.com/nexa_ai/status/2008392901961855074)  2026-01-06T04:19Z [----] followers, [----] engagements


"Your video library just got a brain. 🎞 At CES [----] were introducing Video Search in a new private beta version of Hyperlink. We teamed up with @NVIDIA_AI_PC to ensure Hyperlink runs best on NVIDIA RTX AI PCs using the latest RTX optimizations"  
[X Link](https://x.com/nexa_ai/status/2008591406563278894)  2026-01-06T17:28Z [----] followers, 32.8K engagements


"@hunter_gag37989 Thanks for the support More coming soon"  
[X Link](https://x.com/nexa_ai/status/2008593177616544189)  2026-01-06T17:35Z [----] followers, [--] engagements


"Love the partnership Hyperlinks video understanding and search just clicks on @NVIDIA_AI_PC with RTX acceleration. Join the waitlist for a more natrual way to find the right clip from your local videos: https://hyperlink.nexa.ai/video-rag Welcome to a new way to search. 🔎 Announced at CES [----] we're introducing RTX acceleration for the new @nexa_ai Hyperlink video search capability. Join the waitlist below to get in the beta when it's ready. https://t.co/UIYn6pZQxm https://hyperlink.nexa.ai/video-rag Welcome to a new way to search. 🔎 Announced at CES [----] we're introducing RTX acceleration"  
[X Link](https://x.com/nexa_ai/status/2008610655818182865)  2026-01-06T18:44Z [----] followers, [----] engagements


"On-device AI translation just leveled up 🚀 NexaSDK now supports HY-MT1.5-1.8B @TencentHunyuan's latest 1.8B-parameter multilingual NMT model. Optimized to run directly on the @Qualcomm Hexagon NPU it delivers fast high-quality translation with low latency low power and no cloud dependency. Run with one line of code: nexa infer NexaAI/HY-MT1.5-1.8B-npu Bring real-time multilingual translation to edge devices with NexaSDK https://twitter.com/i/web/status/2008947857064091884 https://twitter.com/i/web/status/2008947857064091884"  
[X Link](https://x.com/nexa_ai/status/2008947857064091884)  2026-01-07T17:04Z [----] followers, [---] engagements


"https://github.com/NexaAI/nexa-sdk https://github.com/NexaAI/nexa-sdk"  
[X Link](https://x.com/nexa_ai/status/2008947859446476840)  2026-01-07T17:04Z [----] followers, [---] engagements


"@memories_ai Nexa AI 🤝 Were partnering to bring visual memory + on-device intelligence to embodied AIso computers cameras glasses and robots can remember and reason locally. With NexaSDK developers can run fast efficient AI inference on local chipsespecially the NPUand reason directly over retrieved memories without sending sensitive data to the cloud. Visual memory + reasoning finally running where the data lives. https://twitter.com/i/web/status/2009309391296815294 https://twitter.com/i/web/status/2009309391296815294"  
[X Link](https://x.com/nexa_ai/status/2009309391296815294)  2026-01-08T17:01Z [----] followers, [---] engagements


"Thank you @AIatAMD for the feature. Seamless PyTorch + ROCm compatibility lowers friction for researchers and accelerates deployment paths for builders. Openness portability and real performance gains are what move the ecosystem forward. Were excited to keep pushing the stack together. Seamless @PyTorch compatibility in ROCm means developers can build once and deploy across both cloud and client. @alexchen4ai CEO of @nexa_ai shares how this simplifies workflows on AMD GPUs. https://t.co/TdT9u2z67R Seamless @PyTorch compatibility in ROCm means developers can build once and deploy across both"  
[X Link](https://x.com/nexa_ai/status/2009446804388237807)  2026-01-09T02:07Z [----] followers, [---] engagements


"CES is a wrap. Huge thanks to @nvidia for the great partnership 🤝 At the NVIDIA booth we demoed Hyperlink's latest multimodal feature: searching videos like text and retrieving the exact clip with precise timestamps powered by @NVIDIA_AI_PC 's RTX-acceleration. The response was immediate. People stopped crowds gathered jaws dropped. Shout-out to @petertheMC for keeping it sharp all week running the demos capturing people's reactions and sharing real-world use cases for creators parents and educators. Join the waitlist to discover moments in your own videos"  
[X Link](https://x.com/nexa_ai/status/2010759403453264348)  2026-01-12T17:02Z [----] followers, [----] engagements


"https://hyperlink.nexa.ai/video-rag https://hyperlink.nexa.ai/video-rag"  
[X Link](https://x.com/nexa_ai/status/2010759405458112957)  2026-01-12T17:02Z [----] followers, [---] engagements


"We just wrapped CES [----] and one of the highlights was demoing Hyperlink at the official @AMD booth. 🤝 Hyperlink is your personal on-device AI super assistant: it understands your files answers with citations and puts @AMDRyzen acceleration to work in real productivitysearch reason generate fully local so its fast + private across creative + technical workflows. The CES reaction honestly blew us away. The signal is clear: this is what the AI PC future looks like. https://twitter.com/i/web/status/2011121473029828821 https://twitter.com/i/web/status/2011121473029828821"  
[X Link](https://x.com/nexa_ai/status/2011121473029828821)  2026-01-13T17:01Z [----] followers, [---] engagements


"We wrapped CES [----] with a strong demo at the official @Qualcomm booth. Hyperlink our on-device AI super assistant performs the latest multimodal agent capabilities on the Qualcomm Hexagon NPU and did two things people immediately felt amazing about: Surfaced and summarized key insights from 1600+ connected files in seconds with inline citations. Understood visual context from an image and added the event straight to Google Calendar with MCP. The demo resonated instantly. Huge thanks to the Qualcomm team for the support. This is what on-device personal intelligence is supposed to feel like."  
[X Link](https://x.com/nexa_ai/status/2011492504697520238)  2026-01-14T17:35Z [----] followers, [---] engagements


"@ultralytics YOLO26 just launched and it already has Day-0 support on @Qualcomm Hexagon NPU with NexaSDK. ⚡ YOLO26 is faster simpler and more efficient for real-world usecrucial for real-time applications like monitoring analytics and safety systems where quick accurate recognition drives smarter decisions and safer automation. With NexaSDK on Hexagon NPU YOLO26 is fully optimized for the edge accelerated by NPUrunning faster using less power and staying entirely on-device across PCs mobile IoT XR and automotive. https://twitter.com/i/web/status/2011842638929100981"  
[X Link](https://x.com/nexa_ai/status/2011842638929100981)  2026-01-15T16:47Z [----] followers, [---] engagements


"Try it today: https://sdk.nexa.ai/model/YOLO26 https://sdk.nexa.ai/model/YOLO26"  
[X Link](https://x.com/nexa_ai/status/2011842641235710230)  2026-01-15T16:47Z [----] followers, [---] engagements


"@ajit_sadalagi @alanzhuly @Qualcomm Great use case Definitely sign up for the On Device Bounty Program"  
[X Link](https://x.com/nexa_ai/status/2011887846265967079)  2026-01-15T19:46Z [----] followers, [--] engagements


"@ajit_sadalagi @alanzhuly @Qualcomm Workshops office hours and meetups will be announced soon to get participants going"  
[X Link](https://x.com/nexa_ai/status/2011900576544829951)  2026-01-15T20:37Z [----] followers, [--] engagements


"https://sdk.nexa.ai/model/LFM2.5-1.2B-Thinking https://sdk.nexa.ai/model/LFM2.5-1.2B-Thinking"  
[X Link](https://x.com/nexa_ai/status/2013657273185497556)  2026-01-20T16:58Z [----] followers, [---] engagements


"@ajit_sadalagi @Qualcomm Let me know if you are able to follow this: https://github.com/NexaAI/nexa-sdk/tree/main/cookbook/android/RAG-LLM https://github.com/NexaAI/nexa-sdk/tree/main/cookbook/android/RAG-LLM"  
[X Link](https://x.com/nexa_ai/status/2014068593697112540)  2026-01-21T20:12Z [----] followers, [--] engagements


"Nexa AI led strong on-device AI momentum at the @HP CES [----] booth. Hyperlink the on-device AI super assistant showcased new agentic workflows that understand local multimodal files reason across context and take action instantly with MCP. Everything runs fully on the @Qualcomm Hexagon NPU on @Snapdragon powered by NexaSDK for industry-leading speed and efficiency. This is next-gen on-device AI built for real work. https://twitter.com/i/web/status/2014748264420294776 https://twitter.com/i/web/status/2014748264420294776"  
[X Link](https://x.com/nexa_ai/status/2014748264420294776)  2026-01-23T17:13Z [----] followers, [--] engagements


"https://hyperlink.nexa.ai/video-rag https://hyperlink.nexa.ai/video-rag"  
[X Link](https://x.com/nexa_ai/status/2015833752099029059)  2026-01-26T17:06Z [----] followers, [---] engagements


"This Week at Nexa: strong momentum across builder community partners and day-0 model support. 1) Nexa @Qualcomm On-Device Bounty Program (Mobile) is live. It's a great chance for builders to ship real @Android AI apps running fully on the Qualcomm Hexagon NPU with NexaSDK and win awards. Register today. 2) Hyperlink lit up CES with three major partners: @Qualcomm booth: Hyperlink demoed the latest multimodal agent capabilities on Qualcomm Hexagon NPU understanding visual context from an image and adding the event straight to Google Calendar with MCP. @nvidia booth Hyperlink unveiled the"  
[X Link](https://x.com/nexa_ai/status/2012209768598040638)  2026-01-16T17:06Z [----] followers, [---] engagements


"NexaSDK now ships Day-0 support for @liquidai latest LFM2.5-1.2B-Thinking on @Qualcomm Hexagon NPU. LFM2.5-1.2B-Thinking delivers best-in-class reasoning quality at sub-2B size while remaining memory-efficient and fast. It enables consistent planning decision-making and multi-step execution ideal for local agentic tasks data extraction and RAG. Benchmarks: - @Snapdragon X Elite NPU (Compute): Prefill speed: [----] tok/s Decode speed: [--] tok/s - Snapdragon [--] Gen [--] NPU (Mobile): Prefill speed: [----] tok/s Decode speed: [--] tok/s - Dragonwing IQ-9075 NPU (IoT): Prefill speed: [----] tok/s Decode"  
[X Link](https://x.com/nexa_ai/status/2013657271621091520)  2026-01-20T16:58Z [----] followers, [----] engagements


"Mobile NPUs can now deliver up to [--] faster time-to-first-token (prefill speed) than the onboard GPU and up to [--] faster than the CPU. Today were shipping a new prefill-speed optimization in NexaML infernece engine that enabled this on @Qualcomm Hexagon NPU. Prefill is when the model reads your prompt/context before generating the first token the core reason why on-device RAG multimodal and agentic apps with long contexts feel slow now. But not anymore Check out the end-to-end mobile RAG demo on @Snapdragon [--] Elite we run the exact same pipeline side-by-side on NPU vs GPU vs CPU. With 1000"  
[X Link](https://x.com/nexa_ai/status/2014029926962438458)  2026-01-21T17:38Z [----] followers, [---] engagements


"Mobile NPUs can now beat computer GPUs on time-to-first-token (TTFT) for long-context local AI. We compared mobile NPU with NexaML's latest prefill-speed optimization on @Qualcomm Hexagon NPU against Apple's flagship M4 Max GPU - same model same quantization same RAG pipeline: Benchmark (Llama 3.2-3B 4-bit prefill): - Apple M4 (GPU): [----] tokens/s - Snapdragon [--] Elite (NPU + NexaML): [----] tokens/s With [----] tokens of context that is 1.5X faster TTFT on a mobile NPU than on a flagship Mac GPU massively eliminating latency for RAG and agents. Your complex laptop-only AI workflows just became"  
[X Link](https://x.com/nexa_ai/status/2014384178612273311)  2026-01-22T17:06Z [----] followers, [----] engagements


"https://github.com/NexaAI/nexa-sdk https://github.com/NexaAI/nexa-sdk"  
[X Link](https://x.com/nexa_ai/status/2014384180529160474)  2026-01-22T17:06Z [----] followers, [---] engagements


"@Qualcomm Excited to see what everyone builds with edge AI"  
[X Link](https://x.com/nexa_ai/status/2014826955208855721)  2026-01-23T22:25Z [----] followers, [--] engagements


"Introducing NexaSDK for Android (Beta) run the latest AI models locally [--] more energy-efficient and [--] faster on @Android devices powered by the @Qualcomm Hexagon NPU. This is the first SDK to support NPU GPU and CPU unlocking the full power of every Android device for example LFM2-1.2B achieves [--] t/s on NPU vs [--] t/s on CPU. With just [--] lines of code you can run the latest state-of-the-art models across every task for example: Multimodal (Vision Audio Text): OmniNeural-4B Embedding: EmbeddingGemma from @Google ASR: Parakeet-v3 from @nvidia OCR: PaddleOCR from @Baidu_Inc Rerank: Jina-reranker"  
[X Link](https://x.com/nexa_ai/status/1986480188813967440)  2025-11-06T17:05Z [----] followers, 26.2K engagements


"Nexa AI led strong on-device AI momentum at the @HP CES [----] booth. Hyperlink the on-device AI super assistant showcased new agentic workflows that understand local multimodal files reason across context and take action instantly with MCP. Everything runs fully on the @Qualcomm Hexagon NPU on @Snapdragon powered by NexaSDK for industry-leading speed and efficiency. This is next-gen on-device AI built for real work. https://twitter.com/i/web/status/2014753976215076984 https://twitter.com/i/web/status/2014753976215076984"  
[X Link](https://x.com/nexa_ai/status/2014753976215076984)  2026-01-23T17:35Z [----] followers, [----] engagements


"This Thursday come hang with us live. We're hosting a Live Q&A on the Qualcomm Developers Discord and YouTube Channel to break down the Nexa AI @Qualcomm On-Device Bounty Program. In [--] minutes we'll cover: Why on-device mobile AI matters right now Prizes timeline and how to join A live demo of what you can build with NexaSDK on Qualcomm @Snapdragon [--] Elite mobile devices Open Q&A bring your questions 📅 February [--] @ [--] AM PT Qualcomm Developers Discord & YouTube Whether you've already submitted a project or you're still exploring this is your chance to talk directly with the team building the"  
[X Link](https://x.com/nexa_ai/status/2018759182602903667)  2026-02-03T18:51Z [----] followers, [--] engagements


"The best vision-language models just went fully on-device - Day-0 on NPU GPU and CPU. Qwen3-VL-4B and 8B from @Alibaba_Qwen now run locally across @Apple @Qualcomm @NVIDIA @Intel @MediaTek and @AMD devices with NexaSDK Every line of model inference code in NexaML GGML and MLX was built from scratch by Nexa for SOTA performance on each hardware stack powered by Nexas unified inference engine. One line to run the latest VLM Day-0 on every backend Qualcomm NPU (NexaML): nexa infer NexaAI/Qwen3-VL-4B-Instruct-NPU nexa infer NexaAI/Qwen3-VL-4B-Thinking-NPU CPU/GPU for everyone (GGML): nexa infer"  
[X Link](https://x.com/nexa_ai/status/1978151146134053085)  2025-10-14T17:29Z [----] followers, 390.9K engagements


"https://sdk.nexa.ai/bounty https://sdk.nexa.ai/bounty"  
[X Link](https://x.com/nexa_ai/status/2016198702810202451)  2026-01-27T17:16Z [----] followers, [---] engagements


"🚀Try LFM2.5-1.2B-Thinking today on @Qualcomm Hexagon with NexaSDK: https://sdk.nexa.ai/model/LFM2.5-1.2B-Thinkingk Today we release LFM2.5-1.2B-Thinking a reasoning model that runs entirely on-device. What needed a data center two years ago now runs on any phone with [---] MB of memory. Trained specifically for concise reasoning Generates internal thinking traces before producing answers https://t.co/dYrwBDefVP https://sdk.nexa.ai/model/LFM2.5-1.2B-Thinkingk Today we release LFM2.5-1.2B-Thinking a reasoning model that runs entirely on-device. What needed a data center two years ago now runs on"  
[X Link](https://x.com/nexa_ai/status/2013663271417483374)  2026-01-20T17:21Z [----] followers, [----] engagements


"Star NexaSDK to stay tuned and download the Demo apk to try: https://github.com/NexaAI/nexa-sdk/tree/main/bindings/android#gpt-oss-npu-demo https://github.com/NexaAI/nexa-sdk/tree/main/bindings/android#gpt-oss-npu-demo"  
[X Link](https://x.com/nexa_ai/status/2016198704022356176)  2026-01-27T17:16Z [----] followers, [---] engagements


"Announcing: Nexa @Qualcomm On-Device Bounty Program Round 1: Mobile On-device AI will be everywhere in [----]. If youre a builder this is your chance to be early and ship something real. Build: A working Android AI app that runs locally on Qualcomm Hexagon NPU (powered by NexaSDK). Youll get: - $6500 total cash prizes - Grand Winner: $5000 cash + Edge AI Impact Award certificate - Top [--] finalists: $500 + flagship Snapdragon powered device - The real upside: Qualcomm marketing spotlight + partnership opportunities plus expert mentorship Timeline (PT): - Jan 15: Launch - Feb 15: Phase [--] deadline"  
[X Link](https://x.com/anyuser/status/2011861038837428488)  2026-01-15T18:00Z [----] followers, 160.2K engagements


"https://github.com/NexaAI/nexa-sdk/tree/main/cookbook/clawd https://github.com/NexaAI/nexa-sdk/tree/main/cookbook/clawd"  
[X Link](https://x.com/anyuser/status/2016570342689615950)  2026-01-28T17:53Z [----] followers, [---] engagements


"https://sdk.nexa.ai/bounty https://sdk.nexa.ai/bounty"  
[X Link](https://x.com/anyuser/status/2017723419157217524)  2026-01-31T22:15Z [----] followers, [---] engagements

Limited data mode. Full metrics available with subscription: lunarcrush.com/pricing