--- title: TornadoVM 7 Enables GPU-Native Workloads From Java... description: TornadoVM, the open-source heterogeneous GPU programming framework for Java, has reached version 7.3. It replaces the Java Native Interface with the Foreign... canonical: https://daily.dev/posts/tornadovm-7-enables-gpu-native-workloads-from-java-applications-sfquydl2z twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: TornadoVM 7 Enables GPU-Native Workloads From Java Applications | daily.dev og:description: TornadoVM, the open-source heterogeneous GPU programming framework for Java, has reached version 7.3. It replaces the Java Native Interface with the Foreign... og:url: https://daily.dev/posts/tornadovm-7-enables-gpu-native-workloads-from-java-applications-sfquydl2z og:image: https://api.daily.dev/og/posts/sFquydl2z.png og:image:alt: TornadoVM 7 Enables GPU-Native Workloads From Java Applications og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # TornadoVM 7 Enables GPU-Native Workloads From Java Applications **[InfoQ](https://daily.dev/sources/infoq)** · 4 min read · 1 upvotes · 0 comments ## Summary TornadoVM, the open-source heterogeneous GPU programming framework for Java, has reached version 7.3. It replaces the Java Native Interface with the Foreign Function & Memory API and drops its experimental JVMCI dependency, sourcing compiler metadata via reflection, ASM and Unsafe across JDK 21, 25, 26 and 27. The release adds a deeper CUDA backend with cuTile support via a new TileContext, direct calls to cuBLAS and CUTLASS, and multi-GPU execution. Alongside it, the team launched jitLLM v2.0, a Java AI inference engine pitched as a competitor to llama.cpp and Ollama, shipped as a single JAR. Both projects are commercialized through Beestream.ai, a spin-off founded by the TornadoVM team. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What changed in TornadoVM 7 compared to earlier versions? TornadoVM 7.3 replaces the Java Native Interface (JNI) with the Foreign Function & Memory (FFM) API in its backends and removes the experimental JVM Compiler Interface (JVMCI) dependency. Compiler metadata is now sourced through a reflection, ASM and Unsafe layer, giving a single uniform path across JDK 21, 25, 26 and 27. The release also adds a deeper CUDA backend, cuTile support via TileContext, direct cuBLAS and CUTLASS calls, and multi-GPU execution. _Java developers tracking GPU acceleration options can follow TornadoVM's architecture changes on daily.dev._ ### Can I call CUDA libraries like cuBLAS directly from Java using TornadoVM? Yes, TornadoVM 7.3 lets developers call CUDA libraries such as cuBLAS and CUTLASS directly from within TornadoVM TaskGraphs via its hybrid API. Developers can swap JIT-generated kernels for these CUDA library calls, or chain the two together, depending on their performance and portability needs. The same hybrid approach is also available for Apple's MLX library. _Java developers evaluating GPU library integration paths can track frameworks like TornadoVM on daily.dev._ ### What is jitLLM and how does it compare to llama.cpp and Ollama? jitLLM is a Java AI inference engine built on TornadoVM, released in version 2.0 alongside TornadoVM 7.3. The TornadoVM team positions it as a direct competitor to llama.cpp and Ollama in performance, and it ships as a single JAR file. Both jitLLM and TornadoVM are commercialized through Beestream.ai, a spin-off founded by the TornadoVM team. _Developers comparing Java-native LLM inference options can follow jitLLM's progress on daily.dev._ ## Similar posts on daily.dev - [TornadoVM 2.0 Brings Automatic GPU Acceleration and LLM support to Java](https://daily.dev/posts/tornadovm-2-0-brings-automatic-gpu-acceleration-and-llm-support-to-java-wdpch8rhk) · InfoQ · 2 upvotes · 0 comments - [TornadoVM 2.0 Released For Java On NVIDIA PTX, OpenCL & SPIR-V Devices](https://daily.dev/posts/tornadovm-2-0-released-for-java-on-nvidia-ptx-opencl-spir-v-devices-il0ezftgh) · Phoronix · 0 upvotes · 0 comments --- Tags: [#java](https://daily.dev/tags/java), [#gpu](https://daily.dev/tags/gpu), [#ai-inference](https://daily.dev/tags/ai-inference), [#cuda](https://daily.dev/tags/cuda) [View this post on daily.dev](https://daily.dev/posts/tornadovm-7-enables-gpu-native-workloads-from-java-applications-sfquydl2z) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"TornadoVM 7 Enables GPU-Native Workloads From Java Applications","url":"https://daily.dev/posts/tornadovm-7-enables-gpu-native-workloads-from-java-applications-sfquydl2z","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/tornadovm-7-enables-gpu-native-workloads-from-java-applications-sfquydl2z"},"datePublished":"2026-10-21T03:00:43.143Z","dateModified":"2026-10-21T03:08:07.421Z","description":"TornadoVM, the open-source heterogeneous GPU programming framework for Java, has reached version 7.3. It replaces the Java Native Interface with the Foreign...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/2499e62f7130a30d8e31a319318c3650?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/2499e62f7130a30d8e31a319318c3650?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"InfoQ","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"InfoQ","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/afc3bced3e1e4b188dd9127017a60e0c","url":"https://daily.dev/sources/infoq"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/tornadovm-7-enables-gpu-native-workloads-from-java-applications-sfquydl2z","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"java,gpu,ai-inference,cuda","timeRequired":"PT4M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"InfoQ","item":"https://daily.dev/sources/infoq"},{"@type":"ListItem","position":3,"name":"TornadoVM 7 Enables GPU-Native Workloads From Java Applications"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/tornadovm-7-enables-gpu-native-workloads-from-java-applications-sfquydl2z#faq","mainEntity":[{"@type":"Question","name":"What changed in TornadoVM 7 compared to earlier versions?","acceptedAnswer":{"@type":"Answer","text":"TornadoVM 7.3 replaces the Java Native Interface (JNI) with the Foreign Function \u0026 Memory (FFM) API in its backends and removes the experimental JVM Compiler Interface (JVMCI) dependency. Compiler metadata is now sourced through a reflection, ASM and Unsafe layer, giving a single uniform path across JDK 21, 25, 26 and 27. The release also adds a deeper CUDA backend, cuTile support via TileContext, direct cuBLAS and CUTLASS calls, and multi-GPU execution. Java developers tracking GPU acceleration options can follow TornadoVM's architecture changes on daily.dev."}},{"@type":"Question","name":"Can I call CUDA libraries like cuBLAS directly from Java using TornadoVM?","acceptedAnswer":{"@type":"Answer","text":"Yes, TornadoVM 7.3 lets developers call CUDA libraries such as cuBLAS and CUTLASS directly from within TornadoVM TaskGraphs via its hybrid API. Developers can swap JIT-generated kernels for these CUDA library calls, or chain the two together, depending on their performance and portability needs. The same hybrid approach is also available for Apple's MLX library. Java developers evaluating GPU library integration paths can track frameworks like TornadoVM on daily.dev."}},{"@type":"Question","name":"What is jitLLM and how does it compare to llama.cpp and Ollama?","acceptedAnswer":{"@type":"Answer","text":"jitLLM is a Java AI inference engine built on TornadoVM, released in version 2.0 alongside TornadoVM 7.3. The TornadoVM team positions it as a direct competitor to llama.cpp and Ollama in performance, and it ships as a single JAR file. Both jitLLM and TornadoVM are commercialized through Beestream.ai, a spin-off founded by the TornadoVM team. Developers comparing Java-native LLM inference options can follow jitLLM's progress on daily.dev."}}]} ```