{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-07-23T07:21:26.498Z","headline":"Thinking Machines Lab 发布开源多模态模型 Inkling","description":"Thinking Machines Lab 发布开源多模态模型 Inkling，总参数 9750 亿（活跃 410 亿），原生支持文本、图像、音频及百万 token 上下文。该模型在 Artificial Analysis Intelligence Index 上以 41 分领跑美国开源模型，但事实准确率仅 40%，幻觉率高达 63%。","url":"https://www.aioga.com/news/cmrncawvl017ubic69x5idyhg/","mainEntityOfPage":"https://www.aioga.com/news/cmrncawvl017ubic69x5idyhg/","datePublished":"2026-07-16T09:55:28.000Z","dateModified":"2026-07-16T09:55:28.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://the-decoder.com/ex-openai-cto-muratis-thinking-machines-drops-inkling-a-975b-parameter-model-that-leads-us-labs-but-trails-china","https://aihot.virxact.com/items/cmrncawvl017ubic69x5idyhg"],"canonicalUrl":"https://www.aioga.com/news/cmrncawvl017ubic69x5idyhg/","directAnswer":{"@type":"Answer","text":"Aioga 编辑摘要：Thinking Machines Lab 发布开源多模态模型 Inkling，总参数 9750 亿（活跃 410 亿），原生支持文本、图像、音频及百万 token 上下文。 Aioga 将其归入「模型更新」方向，重点关注它对真实使用和行业竞争的影响。","url":"https://www.aioga.com/news/cmrncawvl017ubic69x5idyhg/","dateCreated":"2026-07-16T09:55:28.000Z","author":{"@type":"Organization","@id":"https://www.aioga.com/authors/aioga-editorial/#editorial-team","name":"Aioga Editorial Team","url":"https://www.aioga.com/authors/aioga-editorial/"}},"evidence":[{"@type":"CreativeWork","name":"the-decoder.com source article","url":"https://the-decoder.com/ex-openai-cto-muratis-thinking-machines-drops-inkling-a-975b-parameter-model-that-leads-us-labs-but-trails-china","datePublished":"2026-07-16T09:55:28.000Z","provider":{"@type":"Organization","name":"the-decoder.com","url":"https://the-decoder.com/ex-openai-cto-muratis-thinking-machines-drops-inkling-a-975b-parameter-model-that-leads-us-labs-but-trails-china"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.virxact.com/items/cmrncawvl017ubic69x5idyhg","datePublished":"2026-07-16T09:55:28.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.virxact.com/items/cmrncawvl017ubic69x5idyhg"}}],"aggregationSource":"The Decoder：AI News（RSS）","originalPublisher":{"name":"the-decoder.com","url":"https://the-decoder.com/ex-openai-cto-muratis-thinking-machines-drops-inkling-a-975b-parameter-model-that-leads-us-labs-but-trails-china"},"article":{"id":"cmrncawvl017ubic69x5idyhg","slug":"cmrncawvl017ubic69x5idyhg","url":"https://www.aioga.com/news/cmrncawvl017ubic69x5idyhg/","title":"Thinking Machines Lab 发布开源多模态模型 Inkling","title_en":"Ex-OpenAI CTO Murati's Thinking Machines drops Inkling， a 975B parameter model that leads US labs but trails China","summary":"Thinking Machines Lab 发布开源多模态模型 Inkling，总参数 9750 亿（活跃 410 亿），原生支持文本、图像、音频及百万 token 上下文。该模型在 Artificial Analysis Intelligence Index 上以 41 分领跑美国开源模型，但事实准确率仅 40%，幻觉率高达 63%。","source":"The Decoder：AI News（RSS）","sourceUrl":"https://the-decoder.com/ex-openai-cto-muratis-thinking-machines-drops-inkling-a-975b-parameter-model-that-leads-us-labs-but-trails-china","aiHotUrl":"https://aihot.virxact.com/items/cmrncawvl017ubic69x5idyhg","publishedAt":"2026-07-16T09:55:28.000Z","category":"模型更新","score":65,"selected":false,"articleBody":["Thinking Machines Lab, the startup founded by former OpenAI CTO Mira Murati, has released Inkling, an open-weights model with 975 billion parameters. It's built for efficiency and agent-based tasks, but it still trails the best open-source Chinese models in overall performance.","Thinking Machines Lab has shipped its first production-ready language model. Inkling is a Mixture-of-Experts Transformer with 975 billion total parameters, 41 billion of which are active at any given time. It's the first model from the startup founded by Mira Murati, the former OpenAI CTO who played a key role in developing ChatGPT.","Unlike many other open-source AI models, Inkling natively handles text, images, and audio and supports a context window of up to one million tokens. The weights are freely available on Hugging Face：https://huggingface.co/thinkingmachines/inkling. Thinking Machines also offers access through Tinker, its platform for adapting AI models to specific tasks. Ad","The company is positioning Inkling as a flexible base model for customization. \"Inkling is not the strongest overall model available today,\" the announcement：https://thinkingmachines.ai/news/introducing-inkling/ states. Thinking Machines expects the mix of multimodal support, efficient processing, and fine-tuning options to set the model apart. Ad DEC_D_Incontent-1","Thinking Machines says it pre-trained Inkling on 45 trillion tokens of public and synthetic text, images, audio recordings, and videos. The training set also includes public data that \"may be subject to intellectual property protection.\"：https://thinkingmachines.ai/training-data-documentation/ The company used the Chinese AI model Kimi K2.5, among other methods, to generate synthetic data. Kimi K2.5 also served as the basis for Cursor's coding model：https://the-decoder.com/cursor-quietly-built-its-new-coding-model-on-top-of-chinese-open-source-kimi-k2-5/. More technical details are available in the model card：https://thinkingmachines.ai/model-card/inkling/.","According to AI benchmarking platform Artificial Analysis：https://x.com/ArtificialAnlys/status/2077466590346444939, Inkling debuts with a score of 41 on the Artificial Analysis Intelligence Index. That makes it the leading open-weights model from a U.S. lab. It ranks three points above the previous leader, Nemotron 3 Ultra at 38, and well ahead of Gemma 4 31B at 29 and gpt-oss-120b at 24. Ad","On GDPval-AA v2, an agent-based benchmark that simulates knowledge-work tasks, Inkling reaches an Elo rating of 1,238. It beats Kimi K2.6 at 1,190 and DeepSeek v4 Flash max at 1,189. Inkling also scores 24 percent on the Tau-3 banking benchmark, ahead of Kimi K2.6 at 21 percent and DeepSeek v4 Flash max at 23 percent.","Inkling performs rather poorly on factual accuracy. Artificial Analysis gives the model a score of just +2 on its AA Omniscience benchmark. That puts it below the leading open-weights models, though still above other U.S. models such as Nemotron 3 Ultra at -1. Inkling's accuracy is 40 percent, while its hallucination rate is 63 percent. Those results are likely to limit its use in applications that need highly accurate information. Ad DEC_D_Incontent-2","With a 64K context window, Inkling costs $1.87 per million input tokens and $4.68 per million output tokens. That's slightly more than open-source Chinese models such as GLM-5.2 and DeepSeek v4, which offer similar or better performance on text and code tasks. For context windows up to 256,000 tokens, pricing rises to $3.74 for input, $0.748 for cached input, and $9.36 for output. Ad","But Inkling uses fewer output tokens than comparable open-weights models. According to Artificial Analysis, it averages 25,000 output tokens per Intelligence Index task. GLM-5.2 max uses 43,000, Kimi K2.6 uses about 38,000, and DeepSeek v4 Pro max uses about 37,000 tokens on the same tasks.","Thinking Machines says Inkling offers continuously adjustable \"thinking effort.\" Users can choose their preferred balance between cost and performance, reducing token use while maintaining the same result quality.","Thinking Machines is also previewing Inkling-Small：https://thinkingmachines.ai/news/introducing-inkling/#inkling-small, a more compact model with 276 billion total parameters and 12 billion active parameters. The smaller model delivers similar or better results than Inkling on several benchmarks.","Inkling-Small scores 88.3 percent on GPQA Diamond, compared with 87.2 percent for Inkling. On the HLE benchmark with tools, it scores 46.6 percent, slightly ahead of Inkling at 46.0 percent. Thinking Machines credits changes to the pre-training data and training process for the results. The company plans to publish the full weights once testing is complete.","Stay in the loop on AI. Clear, useful, no fluff.","Follow The Decoder for AI news, background stories and expert analyses.","The Decoder：https://the-decoder.com/"],"articleImages":[{"sourceUrl":"https://the-decoder.com/wp-content/uploads/2026/07/inkling_AA_overall_performance-scaled-1.jpg","alt":"Inkling debütiert auf Platz 41 des Artificial Analysis Intelligence Index und ist damit das führende US-Open-Weights-Modell. | Bild: Artificial Analysis","afterParagraph":5,"url":"/media/articles/cmrncawvl017ubic69x5idyhg/3e8894450ada8738.jpg"},{"sourceUrl":"https://the-decoder.com/wp-content/uploads/2026/07/inkling_AA_GDPval_overall_performance-scaled-1.jpg","alt":"","afterParagraph":6,"url":"/media/articles/cmrncawvl017ubic69x5idyhg/2bd5a84c39c3a0ff.jpg"},{"sourceUrl":"https://the-decoder.com/wp-content/uploads/2026/07/inkling_AA_GDPval_token_performance-scaled-1.jpg","alt":"","afterParagraph":7,"url":"/media/articles/cmrncawvl017ubic69x5idyhg/114aa200e068b33c.jpg"}],"mediaStatus":"ok","articleBodyZh":["Thinking Machines Lab，这家由前 OpenAI 首席技术官 Mira Murati 创立的初创公司，发布了 Inkling，一款拥有 9750 亿参数的开源权重模型。它专为高效和基于代理的任务构建，但在整体性能上仍落后于最好的开源中文模型。","Thinking Machines Lab 已推出其首个可以投入生产的语言模型。Inkling 是一款专家混合（Mixture-of-Experts）变换器，总参数量为 9750 亿，其中任意时刻有 410 亿是活跃的。这是由 Mira Murati 创立的初创公司推出的首个模型，她是前 OpenAI 首席技术官，在 ChatGPT 的开发中发挥了关键作用。","与许多其他开源 AI 模型不同，Inkling 原生支持文本、图像和音频，并支持最多一百万个令牌的上下文窗口。权重可以在 Hugging Face 上免费获得：https://huggingface.co/thinkingmachines/inkling。Thinking Machines 还通过 Tinker 平台提供访问，该平台用于将 AI 模型适配到特定任务。广告","该公司将 Inkling 定位为可定制的灵活基础模型。公告中写道：https://thinkingmachines.ai/news/introducing-inkling/，“Inkling 并不是目前可用的最强大的整体模型。”Thinking Machines 预计，其多模态支持、高效处理和微调选项的组合将使模型与众不同。广告 DEC_D_Incontent-1","Thinking Machines 表示，他们在 45 万亿个公共和合成的文本、图像、音频录音及视频上对 Inkling 进行了预训练。训练集还包括“可能受到知识产权保护”的公共数据： https://thinkingmachines.ai/training-data-documentation/。公司使用了中文 AI 模型 Kimi K2.5 等方法生成合成数据。Kimi K2.5 也作为 Cursor 编程模型的基础：https://the-decoder.com/cursor-quietly-built-its-new-coding-model-on-top-of-chinese-open-source-kimi-k2-5/。更多技术细节可在模型卡中查看：https://thinkingmachines.ai/model-card/inkling/。","根据 AI 基准平台 Artificial Analysis（https://x.com/ArtificialAnlys/status/2077466590346444939），Inkling 在 Artificial Analysis 智能指数中首次亮相，得分为 41。这使它成为来自美国实验室的领先开放权重模型。它比之前的领先者 Nemotron 3 Ultra（得分 38）高 3 分，同时远高于 Gemma 4 31B（得分 29）和 gpt-oss-120b（得分 24）。","在 GDPval-AA v2 这一模拟知识工作任务的基于代理的基准测试中，Inkling 的 Elo 评分达到 1,238。它击败了 Kimi K2.6（1,190）和 DeepSeek v4 Flash max（1,189）。在 Tau-3 银行业基准测试中，Inkling 得分为 24%，高于 Kimi K2.6 的 21% 和 DeepSeek v4 Flash max 的 23%。","Inkling 在事实准确性方面表现较差。Artificial Analysis 在其 AA 全知基准测试中给该模型仅打 2 分。这使得它低于领先的开放权重模型，但仍高于其他美国模型，如 Nemotron 3 Ultra（-1）。Inkling 的准确率为 40%，而幻觉率为 63%。这些结果可能限制其在需要高度准确信息的应用中的使用。","在 64K 上下文窗口下，Inkling 的成本为每百万输入 token 1.87 美元，每百万输出 token 4.68 美元。这略高于开源中文模型，如 GLM-5.2 和 DeepSeek v4，在文本和代码任务上提供类似或更好的性能。对于最多 256,000 token 的上下文窗口，定价升至输入 3.74 美元、缓存输入 0.748 美元、输出 9.36 美元。","但 Inkling 使用的输出 token 少于可比的开放权重模型。根据 Artificial Analysis 的数据，它在每个智能指数任务中平均使用 25,000 个输出 token。GLM-5.2 max 使用 43,000 个，Kimi K2.6 使用约 38,000 个，DeepSeek v4 Pro max 使用约 37,000 个 token。","Thinking Machines 表示 Inkling 提供持续可调的“思考努力”。用户可以选择他们偏好的成本与性能之间的平衡，在减少 token 使用的同时保持相同的结果质量。","Thinking Machines 还预览了 Inkling-Small：https://thinkingmachines.ai/news/introducing-inkling/#inkling-small，这是一种更紧凑的模型，具有 2760 亿总参数和 120 亿活跃参数。较小的模型在多个基准测试中表现出与 Inkling 相似或更好的结果。","Inkling-Small 在 GPQA Diamond 上得分为 88.3%，而 Inkling 为 87.2%。在使用工具的 HLE 基准测试中，它得分为 46.6%，略高于 Inkling 的 46.0%。Thinking Machines 将这些结果归因于预训练数据和训练过程的变化。该公司计划在测试完成后发布完整权重。","保持对 AI 的关注。清晰、有用、无废话。","关注 The Decoder，获取 AI 新闻、背景故事和专家分析。","The Decoder：https://the-decoder.com/"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"Aioga 编辑摘要：Thinking Machines Lab 发布开源多模态模型 Inkling，总参数 9750 亿（活跃 410 亿），原生支持文本、图像、音频及百万 token 上下文。 Aioga 将其归入「模型更新」方向，重点关注它对真实使用和行业竞争的影响。","background":"背景分析：模型与研究类动态需要结合能力边界、开放方式、成本、可用性和真实任务表现判断，单项指标领先不等于已经形成稳定采用。","viewpoint":"Aioga 判断：这条动态更适合作为行业观察信号，当前信息足以建立线索，但不足以推导长期结论。","implications":"影响分析：对相关团队而言，短期应先核对来源、可用范围和实际成本，再判断是否值得接入或跟进。","nextStep":"后续观察：继续观察官方文档、实际可用性、价格变化、开发者反馈和竞品回应。","evidenceRefs":["title","summary","articleBody"],"confidence":"medium","status":"published","aiGenerated":false,"autoApproved":true,"generatedBy":"rule-safe-fallback","generatedAt":"2026-07-23T07:30:40.653Z","sourceHash":"1f6c29781e919b75","validation":{"passed":true,"mode":"rule-safe-fallback","checks":["schema","length","source-attribution","no-html"]}},"tags":["模型更新","The Decoder：AI News（RSS）"],"translations":{"zh-CN":{"title":"Thinking Machines Lab 发布开源多模态模型 Inkling","summary":"Thinking Machines Lab 发布开源多模态模型 Inkling，总参数 9750 亿（活跃 410 亿），原生支持文本、图像、音频及百万 token 上下文。该模型在 Artificial Analysis Intelligence Index 上以 41 分领跑美国开源模型，但事实准确率仅 40%，幻觉率高达 63%。","category":"模型更新","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Thinking Machines Lab 发布开源多模态模型 Inkling - Aioga AI资讯","description":"Thinking Machines Lab 发布开源多模态模型 Inkling，总参数 9750 亿（活跃 410 亿），原生支持文本、图像、音频及百万 token 上下文。该模型在 Artificial Analysis Intelligence Index 上以 41 分领跑美国开源模型，但事实准确率仅 40%，幻觉率高达 63%。","url":"https://www.aioga.com/news/cmrncawvl017ubic69x5idyhg/"},"en":{"title":"Thinking Machines Lab releases open-source multimodal model Inkling","summary":"Thinking Machines Lab released the open-source multimodal model Inkling, with a total of 975 billion parameters (41 billion active), natively supporting text, images, audio, and million-token context. The model leads U.S. open-source models on the Artificial Analysis Intelligence Index with a score of 41, but the factual accuracy is only 40%, and the hallucination rate is as high as 63%.","category":"Models","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Thinking Machines Lab releases open-source multimodal model Inkling - Aioga AI News","description":"Thinking Machines Lab released the open-source multimodal model Inkling, with a total of 975 billion parameters (41 billion active), natively supporting text, images, audio, and mi...","url":"https://www.aioga.com/en/news/cmrncawvl017ubic69x5idyhg/","contentTranslated":true,"sourceHash":"e14f9e152229c747","translatedAt":"2026-07-22T19:43:51.365Z"},"ja":{"title":"Thinking Machines Lab、オープンソースのマルチモーダルモデル Inkling を公開","summary":"Thinking Machines LabはオープンソースのマルチモーダルモデルInklingを公開しました。総パラメータは9750億（アクティブは410億）で、テキスト、画像、音声および百万トークンのコンテキストをネイティブにサポートします。このモデルはArtificial Analysis Intelligence Indexで41点を獲得し、アメリカのオープンソースモデルをリードしていますが、事実の正確性はわずか40％で、幻覚率は63％に達します。","category":"モデル更新","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Thinking Machines Lab、オープンソースのマルチモーダルモデル Inkling を公開 - Aioga AIニュース","description":"Thinking Machines LabはオープンソースのマルチモーダルモデルInklingを公開しました。総パラメータは9750億（アクティブは410億）で、テキスト、画像、音声および百万トークンのコンテキストをネイティブにサポートします。このモデルはArtificial Analysis Intelligence Indexで41点を獲得し、アメリカの...","url":"https://www.aioga.com/ja/news/cmrncawvl017ubic69x5idyhg/","contentTranslated":true,"sourceHash":"e14f9e152229c747","translatedAt":"2026-07-22T19:44:00.539Z"},"ko":{"title":"Thinking Machines Lab가 오픈소스 멀티모달 모델 Inkling을 발표","summary":"Thinking Machines Lab는 오픈소스 멀티모달 모델 Inkling을 발표했으며, 총 파라미터는 9750억(활성 410억)이고, 텍스트, 이미지, 오디오 및 백만 토큰 컨텍스트를 기본적으로 지원합니다. 이 모델은 Artificial Analysis Intelligence Index에서 미국 오픈소스 모델을 41점으로 앞섰지만, 사실 정확도는 겨우 40%에 불과하고, 환각률은 63%에 달합니다.","category":"모델 업데이트","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Thinking Machines Lab가 오픈소스 멀티모달 모델 Inkling을 발표 - Aioga AI 뉴스","description":"Thinking Machines Lab는 오픈소스 멀티모달 모델 Inkling을 발표했으며, 총 파라미터는 9750억(활성 410억)이고, 텍스트, 이미지, 오디오 및 백만 토큰 컨텍스트를 기본적으로 지원합니다. 이 모델은 Artificial Analysis Intelligence Index에서 미국 오픈소스 모델을 41...","url":"https://www.aioga.com/ko/news/cmrncawvl017ubic69x5idyhg/","contentTranslated":true,"sourceHash":"e14f9e152229c747","translatedAt":"2026-07-22T19:44:51.910Z"},"es":{"title":"Thinking Machines Lab lanza el modelo multimodal de código abierto Inkling","summary":"Thinking Machines Lab lanzó el modelo multimodal de código abierto Inkling, con un total de 975 mil millones de parámetros (410 mil millones activos), que soporta de manera nativa texto, imagen, audio y contextos de hasta un millón de tokens. Este modelo lidera los modelos de código abierto estadounidenses en el Índice de Inteligencia de Análisis Artificial con 41 puntos, pero la precisión de los hechos es solo del 40% y la tasa de alucinaciones alcanza el 63%.","category":"Modelos","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Thinking Machines Lab lanza el modelo multimodal de código abierto Inkling - Aioga Noticias de IA","description":"Thinking Machines Lab lanzó el modelo multimodal de código abierto Inkling, con un total de 975 mil millones de parámetros (410 mil millones activos), que soporta de manera nativa...","url":"https://www.aioga.com/es/news/cmrncawvl017ubic69x5idyhg/","contentTranslated":true,"sourceHash":"e14f9e152229c747","translatedAt":"2026-07-22T19:44:49.398Z"},"fr":{"title":"Thinking Machines Lab publie le modèle multimodal open source Inkling","summary":"Thinking Machines Lab a publié le modèle multimodal open source Inkling, avec un total de 975 milliards de paramètres (410 milliards actifs), prenant en charge nativement le texte, l'image, l'audio et un contexte de millions de tokens. Ce modèle est en tête des modèles open source américains avec un score de 41 sur l'Artificial Analysis Intelligence Index, mais la précision factuelle n'est que de 40 % et le taux d'hallucination atteint 63 %.","category":"Modèles","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Thinking Machines Lab publie le modèle multimodal open source Inkling - Aioga Actualités IA","description":"Thinking Machines Lab a publié le modèle multimodal open source Inkling, avec un total de 975 milliards de paramètres (410 milliards actifs), prenant en charge nativement le texte,...","url":"https://www.aioga.com/fr/news/cmrncawvl017ubic69x5idyhg/","contentTranslated":true,"sourceHash":"e14f9e152229c747","translatedAt":"2026-07-22T19:45:40.680Z"},"de":{"title":"Thinking Machines Lab veröffentlicht das Open-Source-Multimodalmodell Inkling","summary":"Thinking Machines Lab veröffentlicht das Open-Source-Multimodalmodell Inkling mit insgesamt 975 Milliarden Parametern (davon 41 Milliarden aktiv), das nativ Text, Bilder, Audio und Kontexte mit bis zu einer Million Tokens unterstützt. Dieses Modell führt den amerikanischen Open-Source-Modellindex Artificial Analysis Intelligence Index mit 41 Punkten an, hat jedoch eine tatsächliche Genauigkeit von nur 40 % und eine Halluzinationsrate von bis zu 63 %.","category":"模型更新","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Thinking Machines Lab veröffentlicht das Open-Source-Multimodalmodell Inkling - Aioga KI-News","description":"Thinking Machines Lab veröffentlicht das Open-Source-Multimodalmodell Inkling mit insgesamt 975 Milliarden Parametern (davon 41 Milliarden aktiv), das nativ Text, Bilder, Audio und...","url":"https://www.aioga.com/de/news/cmrncawvl017ubic69x5idyhg/","contentTranslated":true,"sourceHash":"e14f9e152229c747","translatedAt":"2026-07-22T19:45:49.194Z"},"pt-BR":{"title":"Thinking Machines Lab lança modelo multimodal de código aberto Inkling","summary":"O Thinking Machines Lab lançou o modelo multimodal de código aberto Inkling, com um total de 975 bilhões de parâmetros (41 bilhões ativos), suportando nativamente texto, imagem, áudio e contexto de milhões de tokens. Esse modelo lidera os modelos de código aberto dos EUA no Artificial Analysis Intelligence Index com 41 pontos, mas a precisão factual é de apenas 40%, com taxa de alucinação de até 63%.","category":"模型更新","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Thinking Machines Lab lança modelo multimodal de código aberto Inkling - Aioga Notícias de IA","description":"O Thinking Machines Lab lançou o modelo multimodal de código aberto Inkling, com um total de 975 bilhões de parâmetros (41 bilhões ativos), suportando nativamente texto, imagem, áu...","url":"https://www.aioga.com/pt-BR/news/cmrncawvl017ubic69x5idyhg/","contentTranslated":true,"sourceHash":"e14f9e152229c747","translatedAt":"2026-07-22T19:46:38.296Z"},"ru":{"title":"Лаборатория Thinking Machines выпустила открытый мультимодальный модель Inkling","summary":"Thinking Machines Lab выпустила открытое мультимодальное модель Inkling с общим числом параметров 975 миллиардов (активных 41 миллиард), с нативной поддержкой текста, изображений, аудио и контекста на миллионы токенов. Эта модель набрала 41 балл в Artificial Analysis Intelligence Index, лидируя среди американских открытых моделей, но фактическая точность составляет всего 40%, а уровень галлюцинаций достигает 63%.","category":"模型更新","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Лаборатория Thinking Machines выпустила открытый мультимодальный модель Inkling - Aioga Новости ИИ","description":"Thinking Machines Lab выпустила открытое мультимодальное модель Inkling с общим числом параметров 975 миллиардов (активных 41 миллиард), с нативной поддержкой текста, изображений,...","url":"https://www.aioga.com/ru/news/cmrncawvl017ubic69x5idyhg/","contentTranslated":true,"sourceHash":"e14f9e152229c747","translatedAt":"2026-07-22T19:46:40.710Z"},"ar":{"title":"مختبر آلات التفكير يطلق نموذج إنكلين متعدد الوسائط كمصدر مفتوح","summary":"أصدرت مختبرات Thinking Machines نموذج متعدد الوسائط مفتوح المصدر باسم Inkling، بإجمالي 975 مليار معلمة (نشطة 41 مليار)، ويدعم نصوصًا وصورًا وصوتًا وسياقًا يصل إلى مليون وحدة token أصليًا. ويحتل هذا النموذج المرتبة الأولى بين النماذج المفتوحة المصدر الأمريكية في مؤشر الذكاء التحليلي الصناعي برصيد 41، لكن الدقة الواقعية له لا تتجاوز 40%، ومعدل الهلوسة يصل إلى 63%.","category":"模型更新","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"مختبر آلات التفكير يطلق نموذج إنكلين متعدد الوسائط كمصدر مفتوح - Aioga أخبار الذكاء الاصطناعي","description":"أصدرت مختبرات Thinking Machines نموذج متعدد الوسائط مفتوح المصدر باسم Inkling، بإجمالي 975 مليار معلمة (نشطة 41 مليار)، ويدعم نصوصًا وصورًا وصوتًا وسياقًا يصل إلى مليون وحدة token...","url":"https://www.aioga.com/ar/news/cmrncawvl017ubic69x5idyhg/","contentTranslated":true,"sourceHash":"e14f9e152229c747","translatedAt":"2026-07-22T19:47:35.432Z"},"hi":{"title":"Thinking Machines Lab ने ओपन-सोर्स मल्टीमॉडल मॉडल Inkling जारी किया","summary":"Thinking Machines Lab ने ओपन-सोर्स मल्टीमॉडल मॉडल Inkling जारी किया, जिसमें कुल पैरामीटर 9750 अरब (सक्रिय 410 अरब) हैं, और यह मूल रूप से टेक्स्ट, इमेज, ऑडियो और लाखों टोकन संदर्भ का समर्थन करता है। यह मॉडल Artificial Analysis Intelligence Index पर 41 अंकों के साथ अमेरिकी ओपन-सोर्स मॉडल में सबसे आगे है, लेकिन इसका तथ्यात्मक सटीकता केवल 40% है, और भ्रम दर 63% तक है।","category":"模型更新","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Thinking Machines Lab ने ओपन-सोर्स मल्टीमॉडल मॉडल Inkling जारी किया - Aioga AI समाचार","description":"Thinking Machines Lab ने ओपन-सोर्स मल्टीमॉडल मॉडल Inkling जारी किया, जिसमें कुल पैरामीटर 9750 अरब (सक्रिय 410 अरब) हैं, और यह मूल रूप से टेक्स्ट, इमेज, ऑडियो और लाखों टोकन संदर्भ क...","url":"https://www.aioga.com/hi/news/cmrncawvl017ubic69x5idyhg/","contentTranslated":true,"sourceHash":"e14f9e152229c747","translatedAt":"2026-07-22T19:47:38.059Z"},"it":{"title":"Thinking Machines Lab rilascia il modello multimodale open source Inkling","summary":"Thinking Machines Lab ha rilasciato il modello multimodale open source Inkling, con un totale di 975 miliardi di parametri (attivi 41 miliardi), che supporta nativamente testo, immagini, audio e contesti fino a milioni di token. Questo modello guida i modelli open source statunitensi sull'Artificial Analysis Intelligence Index con 41 punti, ma la precisione dei fatti è solo del 40% e il tasso di allucinazioni arriva al 63%.","category":"模型更新","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Thinking Machines Lab rilascia il modello multimodale open source Inkling - Aioga Notizie IA","description":"Thinking Machines Lab ha rilasciato il modello multimodale open source Inkling, con un totale di 975 miliardi di parametri (attivi 41 miliardi), che supporta nativamente testo, imm...","url":"https://www.aioga.com/it/news/cmrncawvl017ubic69x5idyhg/","contentTranslated":true,"sourceHash":"e14f9e152229c747","translatedAt":"2026-07-22T19:48:30.345Z"},"nl":{"title":"Thinking Machines Lab brengt het open-source multimodale model Inkling uit","summary":"Thinking Machines Lab heeft het open-source multimodale model Inkling uitgebracht, met in totaal 975 miljard parameters (actief 41 miljard), en ondersteunt van nature tekst, afbeeldingen, audio en contexten van miljoenen tokens. Dit model staat met 41 punten bovenaan de Amerikaanse open-source modellen op de Artificial Analysis Intelligence Index, maar de feitelijke nauwkeurigheid is slechts 40% en de hallucinatiegraad is tot 63%.","category":"模型更新","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Thinking Machines Lab brengt het open-source multimodale model Inkling uit - Aioga AI-nieuws","description":"Thinking Machines Lab heeft het open-source multimodale model Inkling uitgebracht, met in totaal 975 miljard parameters (actief 41 miljard), en ondersteunt van nature tekst, afbeel...","url":"https://www.aioga.com/nl/news/cmrncawvl017ubic69x5idyhg/","contentTranslated":true,"sourceHash":"e14f9e152229c747","translatedAt":"2026-07-22T19:48:27.722Z"},"tr":{"title":"Thinking Machines Lab, açık kaynaklı çok modlu model Inkling'i yayınladı","summary":"Thinking Machines Lab, açık kaynaklı çok modlu model Inkling’i yayınladı; toplam parametre sayısı 975 milyar (aktif 41 milyar) olup, metin, görüntü, ses ve milyon token bağlamını yerel olarak desteklemektedir. Bu model, Artificial Analysis Intelligence Index’te 41 puanla ABD açık kaynak modellerini geride bırakıyor, ancak gerçek doğruluk oranı yalnızca %40 ve halüsinasyon oranı %63.","category":"模型更新","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Thinking Machines Lab, açık kaynaklı çok modlu model Inkling'i yayınladı - Aioga AI Haberleri","description":"Thinking Machines Lab, açık kaynaklı çok modlu model Inkling’i yayınladı; toplam parametre sayısı 975 milyar (aktif 41 milyar) olup, metin, görüntü, ses ve milyon token bağlamını y...","url":"https://www.aioga.com/tr/news/cmrncawvl017ubic69x5idyhg/","contentTranslated":true,"sourceHash":"e14f9e152229c747","translatedAt":"2026-07-22T19:49:18.432Z"},"vi":{"title":"Thinking Machines Lab phát hành mô hình đa phương thức mã nguồn mở Inkling","summary":"Thinking Machines Lab phát hành mô hình đa phương thức nguồn mở Inkling, tổng số tham số 975 tỷ (hoạt động 41 tỷ), hỗ trợ gốc văn bản, hình ảnh, âm thanh và ngữ cảnh lên đến hàng triệu token. Mô hình này dẫn đầu các mô hình nguồn mở của Mỹ trên Chỉ số Trí tuệ Phân tích Nhân tạo với 41 điểm, nhưng độ chính xác thực tế chỉ đạt 40%, tỉ lệ ảo tưởng lên tới 63%.","category":"模型更新","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Thinking Machines Lab phát hành mô hình đa phương thức mã nguồn mở Inkling - Tin tức AI Aioga","description":"Thinking Machines Lab phát hành mô hình đa phương thức nguồn mở Inkling, tổng số tham số 975 tỷ (hoạt động 41 tỷ), hỗ trợ gốc văn bản, hình ảnh, âm thanh và ngữ cảnh lên đến hàng t...","url":"https://www.aioga.com/vi/news/cmrncawvl017ubic69x5idyhg/","contentTranslated":true,"sourceHash":"e14f9e152229c747","translatedAt":"2026-07-22T19:49:28.519Z"},"id":{"title":"Thinking Machines Lab merilis model multimodal open source Inkling","summary":"Thinking Machines Lab merilis model multimodal open-source Inkling, dengan total parameter 975 miliar (aktif 41 miliar), mendukung secara native teks, gambar, audio, dan konteks hingga jutaan token. Model ini memimpin model open-source Amerika dalam Artificial Analysis Intelligence Index dengan skor 41, tetapi akurasi faktual hanya 40%, dan tingkat halusinasi mencapai 63%.","category":"模型更新","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Thinking Machines Lab merilis model multimodal open source Inkling - Berita AI Aioga","description":"Thinking Machines Lab merilis model multimodal open-source Inkling, dengan total parameter 975 miliar (aktif 41 miliar), mendukung secara native teks, gambar, audio, dan konteks hi...","url":"https://www.aioga.com/id/news/cmrncawvl017ubic69x5idyhg/","contentTranslated":true,"sourceHash":"e14f9e152229c747","translatedAt":"2026-07-22T19:50:09.761Z"},"th":{"title":"ห้องแล็บ Thinking Machines เปิดตัวโมเดลมัลติโมดัล Inkling แบบโอเพนซอร์ส","summary":"Thinking Machines Lab ได้เปิดตัวโมเดลมัลติโมดัลแบบโอเพนซอร์สชื่อ Inkling โดยมีพารามิเตอร์ทั้งหมด 975 พันล้าน (พารามิเตอร์ใช้งานจริง 41 พันล้าน) รองรับข้อความ รูปภาพ เสียง และบริบทล้านโทเค็นโดยตรง โมเดลนี้ทำคะแนนนำโมเดลโอเพนซอร์สของสหรัฐฯ ด้วย 41 คะแนนบน Artificial Analysis Intelligence Index แต่ความถูกต้องของข้อเท็จจริงอยู่ที่ 40% และอัตราการหลอกลวงสูงถึง 63%","category":"模型更新","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"ห้องแล็บ Thinking Machines เปิดตัวโมเดลมัลติโมดัล Inkling แบบโอเพนซอร์ส - ข่าว AI Aioga","description":"Thinking Machines Lab ได้เปิดตัวโมเดลมัลติโมดัลแบบโอเพนซอร์สชื่อ Inkling โดยมีพารามิเตอร์ทั้งหมด 975 พันล้าน (พารามิเตอร์ใช้งานจริง 41 พันล้าน) รองรับข้อความ รูปภาพ เสียง และบริบทล...","url":"https://www.aioga.com/th/news/cmrncawvl017ubic69x5idyhg/","contentTranslated":true,"sourceHash":"e14f9e152229c747","translatedAt":"2026-07-22T19:50:23.644Z"},"pl":{"title":"Thinking Machines Lab wydaje otwarty model multimodalny Inkling","summary":"Thinking Machines Lab opublikowało otwartoźródłowy model multimodalny Inkling, z całkowitą liczbą parametrów 975 miliardów (aktywnych 41 miliardów), natywnie obsługujący tekst, obraz, dźwięk oraz kontekst o długości miliona tokenów. Model ten w Artificial Analysis Intelligence Index zdobył 41 punktów, prowadząc wśród amerykańskich modeli open source, ale dokładność faktów wynosi tylko 40%, a współczynnik halucynacji aż 63%.","category":"模型更新","source":"The Decoder：AI News（RSS）","aggregationSource":"The Decoder：AI News（RSS）","pageTitle":"Thinking Machines Lab wydaje otwarty model multimodalny Inkling - Aioga Wiadomości AI","description":"Thinking Machines Lab opublikowało otwartoźródłowy model multimodalny Inkling, z całkowitą liczbą parametrów 975 miliardów (aktywnych 41 miliardów), natywnie obsługujący tekst, obr...","url":"https://www.aioga.com/pl/news/cmrncawvl017ubic69x5idyhg/","contentTranslated":true,"sourceHash":"e14f9e152229c747","translatedAt":"2026-07-22T19:51:13.420Z"}}}}