←Search results

English text is being prepared.

By relevance · Total 121 results Clear search

Friday

  1. FactRelated story

    NVIDIA releases on-device small model Nemotron-3.5-Bolt-E4B

    NVIDIA releases the on-device small model Nemotron-3.5-Bolt-E4B. A report on October 23, 2026 states that the model can run locally on devices with 6GB of memory, targeting on-device devices such as phones; it is trained based on Gemma 4, supports vision, and Unsloth simultaneously provides the GGUF file and a guide link. NVIDIA says it targets scenarios such as local assistants, interactive characters, and robots that can converse and call tools. A follow-up report on October 24 states that the model has 8B total parameters, 4B effective parameters, and a 64K context, and is post-trained using only reinforcement learning based on Google Gemma 4 E4B; the Q4_K_M version occupies 3.1 to 4.3 GB of VRAM on RTX and can run on GeForce RTX, Jetson, and DGX Spark. The report also states that in NVIDIA's own testing, it leads Gemma 4 E4B, Qwen3.5-4B, and Nemotron 3 Nano 4B in 15 of 21 benchmarks, but Tau2-Bench is 55.1, trailing Qwen3.5-4B's 75.1.

    Matched fieldsTitle

    View story:NVIDIA releases on-device small model Nemotron-3.5-Bolt-E4B

Sunday

  1. Story1 reports · 1 facts

    Meta distills a 30B on-device agent model, Muse Glimmer

    Meta distilled the dense 30B model Muse Glimmer from the larger Muse Spark, training it on synthetic agent trajectories in simulated applications. Its 4-bit weights are under 20 GB, allowing it to run on a single RTX 5090. According to a report dated October 22, 2026, on the multi-tool benchmark MCPAtlas, Muse Glimmer scored 75.5, higher than Gemma4-31B's 54.2 and Qwen3.6-27B's 62.5; however, on OSWorld-Verified it lagged behind Qwen3.6-27B's 75.6 with a score of 65.9. There are currently no subsequent developments.

    Meta distilled the dense 30B model Muse Glimmer from the larger Muse Spark, training it on synthetic agent trajectories in simulated applications. Its 4-bit weights are under 20 GB, allowing it to run on a single RTX 5090. According to a report dated October 22, 2026, on the multi-tool benchmark MCPAtlas, Muse Glimmer scored 75.5, higher than Gemma4-31B's 54.2 and Qwen3.6-27B's 62.5; however, on OSWorld-Verified it lagged behind Qwen3.6-27B's 75.6 with a score of 65.9. There are currently no sub

    Matched fieldsTitle

  2. Story1 reports · 1 facts

    CAICT and the AIIA On-Device Intelligence Working Group Jointly Release the On-Device Intelligence Industry Development Report

    On October 14, 2026, the Artificial Intelligence Research Institute of the China Academy of Information and Communications Technology (CAICT), together with the On-Device Intelligence Working Group of the Artificial Intelligence Industry Alliance (AIIA), and Huawei, ZTE, vivo, Haier, StepFun, ModelBest, Dahua Technology and other organizations released the "On-Device Intelligence Industry Development Report (2026)." The report defines on-device intelligence as a primarily local, device-cloud collaborative form of integrated intelligence, focusing on key technologies such as model lightweighting, heterogeneous hardware integration, foundational software, agents, device-cloud collaboration, and security and trustworthiness. The report also outlines native product forms such as personal intelligence, spatial intelligence, mobile intelligence, industry intelligence nodes, and embodied intelligence, as well as application scenarios including consumer entertainment, smart homes, smart offices, smart mobility, and smart health. The report has now been publicly released, and no further developments have been disclosed.

    On October 14, 2026, the Artificial Intelligence Research Institute of the China Academy of Information and Communications Technology (CAICT), together with the On-Device Intelligence Working Group of the Artificial Intelligence Industry Alliance (AIIA), and Huawei, ZTE, vivo, Haier, StepFun, ModelBest, Dahua Technology and other organizations released the "On-Device Intelligence Industry Development Report (2026)." The report defines on-device intelligence as a primarily local, device-cloud col

    Matched fieldsTitle

  3. Story1 reports · 1 facts

    OpenBMB showcases MiniCPM-V 4.7 1B running on-device on an ordinary iPhone

    OpenBMB (ModelBest) showcased MiniCPM-V 4.7 1B running locally on an ordinary iPhone via OnDeviceLLM. Related reports say the model is no longer limited to the 12GB-memory Pro Max models, and can complete vision tasks such as reading receipts and calculating totals, extracting flyer information, and understanding screenshots offline on-device, and can also be used in airplane mode. Currently this information comes from the same occurrence evidence from a single source, with no subsequent independent developments yet.

    OpenBMB (ModelBest) showcased MiniCPM-V 4.7 1B running locally on an ordinary iPhone via OnDeviceLLM. Related reports say the model is no longer limited to the 12GB-memory Pro Max models, and can complete vision tasks such as reading receipts and calculating totals, extracting flyer information, and understanding screenshots offline on-device, and can also be used in airplane mode. Currently this information comes from the same occurrence evidence from a single source, with no subsequent indepen

    Matched fieldsTitle

  4. Story1 reports · 1 facts

    Matt Turck says open-source voice AI welcomes a small on-device SOTA TTS model suite

    Matt Turck reposted and commented that a small on-device SOTA text-to-speech model suite has emerged in the open-source voice AI space. According to related reports, Gradium has open-sourced the Phonon on-device speech synthesis model suite, with all model parameters under 50m, focusing on multilingual naturalness and accuracy, and claiming to offer the broadest inference support to date; among them, Phonon-31m is released under the Apache 2.0 license. Currently, this news has been reposted by Matt Turck, with no further developments yet.

    Matt Turck reposted and commented that a small on-device SOTA text-to-speech model suite has emerged in the open-source voice AI space. According to related reports, Gradium has open-sourced the Phonon on-device speech synthesis model suite, with all model parameters under 50m, focusing on multilingual naturalness and accuracy, and claiming to offer the broadest inference support to date; among them, Phonon-31m is released under the Apache 2.0 license. Currently, this news has been reposted by M

    Matched fieldsTitle

  5. FactRelated story

    Matt Turck says open-source voice AI welcomes a small on-device SOTA TTS model suite

    Matt Turck reposted and commented that a small on-device SOTA text-to-speech model suite has emerged in the open-source voice AI space. According to related reports, Gradium has open-sourced the Phonon on-device speech synthesis model suite, with all model parameters under 50m, focusing on multilingual naturalness and accuracy, and claiming to offer the broadest inference support to date; among them, Phonon-31m is released under the Apache 2.0 license. Currently, this news has been reposted by Matt Turck, with no further developments yet.

    Matched fieldsTitle

    View story:Matt Turck says open-source voice AI welcomes a small on-device SOTA TTS model suite

  6. Story1 reports · 1 facts

    Meta compresses its Agent model Muse Glimmer 30B to run on-device

    On October 20, 2026, Meta released the technical report for Muse Glimmer, a 30B open-weight model aimed at agent tasks on consumer-grade hardware, running at 4-bit precision on a single consumer GPU. The reporter adds that the model leads open-source models of the same scale in tool calling and multi-step planning, surpassing Qwen3.6-27B and Gemma4-31B on some metrics; after 4-bit quantization it is about 17GB, targeting devices with 24GB of VRAM, and uses DFlash speculative decoding to reach about 233 tokens/s on an RTX 5090, roughly 3.1 times faster than ordinary decoding. There are currently no subsequent developments.

    On October 20, 2026, Meta released the technical report for Muse Glimmer, a 30B open-weight model aimed at agent tasks on consumer-grade hardware, running at 4-bit precision on a single consumer GPU. The reporter adds that the model leads open-source models of the same scale in tool calling and multi-step planning, surpassing Qwen3.6-27B and Gemma4-31B on some metrics; after 4-bit quantization it is about 17GB, targeting devices with 24GB of VRAM, and uses DFlash speculative decoding to reach ab

    Matched fieldsTitle

Thursday

  1. FactRelated story

    Meta distills a 30B on-device agent model, Muse Glimmer

    Meta distilled the dense 30B model Muse Glimmer from the larger Muse Spark, training it on synthetic agent trajectories in simulated applications. Its 4-bit weights are under 20 GB, allowing it to run on a single RTX 5090. According to a report dated October 22, 2026, on the multi-tool benchmark MCPAtlas, Muse Glimmer scored 75.5, higher than Gemma4-31B's 54.2 and Qwen3.6-27B's 62.5; however, on OSWorld-Verified it lagged behind Qwen3.6-27B's 75.6 with a score of 65.9. There are currently no subsequent developments.

    Matched fieldsTitle

    View story:Meta distills a 30B on-device agent model, Muse Glimmer

Wednesday

  1. FactRelated story

    CAICT and the AIIA On-Device Intelligence Group jointly release the On-Device Intelligence Industry Development Report

    On October 14, 2026, the Artificial Intelligence Research Institute of the China Academy of Information and Communications Technology (CAICT), together with the On-Device Intelligence Working Group of the Artificial Intelligence Industry Alliance (AIIA), and Huawei, ZTE, vivo, Haier, StepFun, ModelBest, Dahua Technology and other organizations released the "On-Device Intelligence Industry Development Report (2026)." The report defines on-device intelligence as a primarily local, device-cloud collaborative form of integrated intelligence, focusing on key technologies such as model lightweighting, heterogeneous hardware integration, foundational software, agents, device-cloud collaboration, and security and trustworthiness. The report also outlines native product forms such as personal intelligence, spatial intelligence, mobile intelligence, industry intelligence nodes, and embodied intelligence, as well as application scenarios including consumer entertainment, smart homes, smart offices, smart mobility, and smart health. The report has now been publicly released, and no further developments have been disclosed.

    Matched fieldsTitle

    View story:CAICT and the AIIA On-Device Intelligence Working Group Jointly Release the On-Device Intelligence Industry Development Report

  6. Story3 reports · 2 facts

    Liquid AI Releases LFM2.5-24B-A2B On-Device Reasoning Model

    On October 13, 2026, Liquid AI released the LFM2.5-24B-A2B on-device agent model. With 24B total parameters and only 2B activated per token, it is the largest and final model in the LFM2.5 family, and both the base and post-trained versions have open weights on Hugging Face and Playground. According to the official blog, it uses a pure reasoning mode with explicit chain-of-thought, expands the context window from 32,768 to 131,072 tokens, grows the vocabulary from 65K to 128K, and scales pretraining data from 17T to 33T tokens; it leads or ties Qwen3.6-35B-A3B on five of eight agent and tool-calling benchmarks, scoring 27.69 versus 8.72 on AutomationBench and 48.09 versus 38.19 on LiveClaw; it decodes at 184 tokens/s on an Apple M5 Max and outputs over 21K tokens/s under high concurrency on a single NVIDIA B300 GPU. On October 14, OpenRouter announced that the model is now available, saying it reasons before answering and targets agent and tool-calling scenarios.

    On October 13, 2026, Liquid AI released the LFM2.5-24B-A2B on-device agent model. With 24B total parameters and only 2B activated per token, it is the largest and final model in the LFM2.5 family, and both the base and post-trained versions have open weights on Hugging Face and Playground. According to the official blog, it uses a pure reasoning mode with explicit chain-of-thought, expands the context window from 32,768 to 131,072 tokens, grows the vocabulary from 65K to 128K, and scales pretrai

    Matched fieldsTitle

Wednesday

  1. FactRelated story

    OpenBMB demonstrates MiniCPM-V 4.7 1B running on-device on iPhone

    OpenBMB (ModelBest) showcased MiniCPM-V 4.7 1B running locally on an ordinary iPhone via OnDeviceLLM. Related reports say the model is no longer limited to the 12GB-memory Pro Max models, and can complete vision tasks such as reading receipts and calculating totals, extracting flyer information, and understanding screenshots offline on-device, and can also be used in airplane mode. Currently this information comes from the same occurrence evidence from a single source, with no subsequent independent developments yet.

    Matched fieldsTitle

    View story:OpenBMB showcases MiniCPM-V 4.7 1B running on-device on an ordinary iPhone

Tuesday

  1. FactRelated story

    Meta compresses its agent model Muse Glimmer 30B to run on-device

    On October 20, 2026, Meta released the technical report for Muse Glimmer, a 30B open-weight model aimed at agent tasks on consumer-grade hardware, running at 4-bit precision on a single consumer GPU. The reporter adds that the model leads open-source models of the same scale in tool calling and multi-step planning, surpassing Qwen3.6-27B and Gemma4-31B on some metrics; after 4-bit quantization it is about 17GB, targeting devices with 24GB of VRAM, and uses DFlash speculative decoding to reach about 233 tokens/s on an RTX 5090, roughly 3.1 times faster than ordinary decoding. There are currently no subsequent developments.

    Matched fieldsTitle

    View story:Meta compresses its Agent model Muse Glimmer 30B to run on-device

Friday

Thursday

Tuesday

  1. FactRelated story

    English translation is being prepared.

    On October 13, 2026, Liquid AI released the LFM2.5-24B-A2B on-device agent model. With 24B total parameters and only 2B activated per token, it is the largest and final model in the LFM2.5 family, and both the base and post-trained versions have open weights on Hugging Face and Playground. According to the official blog, it uses a pure reasoning mode with explicit chain-of-thought, expands the context window from 32,768 to 131,072 tokens, grows the vocabulary from 65K to 128K, and scales pretraining data from 17T to 33T tokens; it leads or ties Qwen3.6-35B-A3B on five of eight agent and tool-calling benchmarks, scoring 27.69 versus 8.72 on AutomationBench and 48.09 versus 38.19 on LiveClaw; it decodes at 184 tokens/s on an Apple M5 Max and outputs over 21K tokens/s under high concurrency on a single NVIDIA B300 GPU. On October 14, OpenRouter announced that the model is now available, saying it reasons before answering and targets agent and tool-calling scenarios.

    Matched fieldsTitle

    View story:Liquid AI Releases LFM2.5-24B-A2B On-Device Reasoning Model