{"content_id":"onwrq0m9id","slug":"google-gemini-robotics-2-5-partner-rollout","locale":"en","schema_type":"NewsArticle","category":"news","category_name":"News","title":"Google Gemini Robotics 2.5 rolls out to partners","summary":"Google DeepMind is rolling out Gemini Robotics 2.5 to robotics partners. Availability in Korea, the public API release schedule for general developers, and pricing have not been disclosed.","sponsorship_disclosure":null,"affiliate_disclosure":null,"commerce_disclosure":null,"author":{"name":"Injoys Editorial Team","url":"https://injoys.com/ko/about"},"key_points":["Gemini Robotics 2.5 is a VLA model that converts visual and language inputs into robot motor control.","Google DeepMind said it is rolling out the model to robotics partners.","Google reported that Apollo 2 achieved a 96% task success rate in a proof of concept for picking and packing automotive parts.","Google said its own evaluations showed improvements in task success rates and completion speed over the previous baseline model.","Availability in Korea, the public API release schedule for general developers, pricing, and usage limits have not been disclosed."],"content_markdown":"Google DeepMind has announced Gemini Robotics 2.5. The announcement was published on November 10, 2026. It is currently rolling out to **robotics partners**. Availability to general developers in Korea has not been disclosed.\n\nGoogle DeepMind's announcement and model description were reviewed as of November 12, 2026.\n\n## A model that turns visual and language inputs into robot actions\n\nGemini Robotics 2.5 is a vision-language-action model. Abbreviated as VLA, it converts visual and language inputs into motor control. Its role is to enable robots to take physical actions based on those inputs.\n\nGoogle says it can control robots ranging from tabletop robots to humanoids. This is the model's control scope as described by the company. The terms of use must be checked separately when applying it to individual robots. [Google DeepMind model description](https://deepmind.google/models/gemini-robotics/vla/)\n\n## Rolling out to partners, with availability in Korea undisclosed\n\nGoogle's robotics partners are the only recipients currently specified. Google DeepMind said it is rolling out the model to these partners. It did not announce a timeline for a public API for general developers.\n\nAvailability in Korea and regional coverage have not been disclosed either. The announcement alone therefore does not establish whether developers in Korea can use it immediately. The model ID needed for API calls has not been confirmed either.\n\n## Pricing and usage limits undisclosed\n\nPricing and usage limits were not disclosed in the official announcement. There is no basis for calculating costs from pricing plans or API call limits. The news of partner access should not be interpreted as an announcement of free access.\n\n## Company evaluations report improvements in success rates and task speed\n\nThe published performance figures are results reported by Google. They include proof-of-concept results from an automotive production line. The company also announced its own evaluations against the previous baseline model.\n\n| Evaluation metric | Company-reported figure | Scope and conditions |\n| --- | --- | --- |\n| Success rate for picking and packing automotive parts | 96% | Apptronik Apollo 2 proof of concept on a Mercedes-Benz production line |\n| Improvement in task success rate | 30~100% higher than the previous baseline model | Google's own evaluation, with differences by task type |\n| Improvement in task completion speed | 20~30% faster than the previous baseline model | Results against the previous baseline model in Google's own evaluation |\n\nThese figures appeared in the November 10, 2026 announcement. The 96% success rate applies to picking and packing automotive parts. It is not a success rate that applies to all robots and all tasks.\n\nThe improvement in success rate is a relative figure measured against a comparison baseline. It should not be read as an increase of 30~100 percentage points. The improvement in task completion speed must also be distinguished from a reduction in time taken. [Google DeepMind announcement](https://deepmind.google/blog/gemini-robotics-2-5-accelerates-real-world-physical-intelligence/)\n\n## Migration procedures and independent performance measurements not confirmed\n\nOfficial sources did not specify migration procedures for users of existing models. Whether existing access permissions allow use of the new model has not been confirmed either. Anyone preparing to adopt it should check the migration requirements separately.\n\nNo independent performance measurements from Artificial Analysis were presented either. The company's own evaluations must be distinguished from independent measurements. There is no basis for labeling the figures currently presented as results verified by an independent organization.\n\n## Terms to check before adoption in Korea\n\nAnyone considering adoption in Korea should first check whether it is available there. Access for partners must be distinguished from a public API for general developers. The following items should be checked before assessing use or costs.\n\n- Whether it is available in Korea and whether you qualify for partner access\n- The release timeline for a public API for general developers and the exact model ID\n- Applicable pricing and usage limits\n- Procedures required to migrate from an existing model\n\nThe sources to check are Google DeepMind's official announcement and model information page. Terms that are not currently public must be checked in subsequent official updates. When reviewing performance, the task conditions in the company's evaluations also need to be checked.","content_html":"\u003cp\u003eGoogle DeepMind has announced Gemini Robotics 2.5. The announcement was published on November 10, 2026. It is currently rolling out to \u003cstrong\u003erobotics partners\u003c/strong\u003e. Availability to general developers in Korea has not been disclosed.\u003c/p\u003e\n\u003cp\u003eGoogle DeepMind's announcement and model description were reviewed as of November 12, 2026.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#a-model-that-turns-visual-and-language-inputs-into-robot-actions\" class=\"anchor\" id=\"a-model-that-turns-visual-and-language-inputs-into-robot-actions\"\u003e\u003c/a\u003eA model that turns visual and language inputs into robot actions\u003c/h2\u003e\n\u003cp\u003eGemini Robotics 2.5 is a vision-language-action model. Abbreviated as VLA, it converts visual and language inputs into motor control. Its role is to enable robots to take physical actions based on those inputs.\u003c/p\u003e\n\u003cp\u003eGoogle says it can control robots ranging from tabletop robots to humanoids. This is the model's control scope as described by the company. The terms of use must be checked separately when applying it to individual robots. \u003ca href=\"https://deepmind.google/models/gemini-robotics/vla/\"\u003eGoogle DeepMind model description\u003c/a\u003e\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#rolling-out-to-partners-with-availability-in-korea-undisclosed\" class=\"anchor\" id=\"rolling-out-to-partners-with-availability-in-korea-undisclosed\"\u003e\u003c/a\u003eRolling out to partners, with availability in Korea undisclosed\u003c/h2\u003e\n\u003cp\u003eGoogle's robotics partners are the only recipients currently specified. Google DeepMind said it is rolling out the model to these partners. It did not announce a timeline for a public API for general developers.\u003c/p\u003e\n\u003cp\u003eAvailability in Korea and regional coverage have not been disclosed either. The announcement alone therefore does not establish whether developers in Korea can use it immediately. The model ID needed for API calls has not been confirmed either.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#pricing-and-usage-limits-undisclosed\" class=\"anchor\" id=\"pricing-and-usage-limits-undisclosed\"\u003e\u003c/a\u003ePricing and usage limits undisclosed\u003c/h2\u003e\n\u003cp\u003ePricing and usage limits were not disclosed in the official announcement. There is no basis for calculating costs from pricing plans or API call limits. The news of partner access should not be interpreted as an announcement of free access.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#company-evaluations-report-improvements-in-success-rates-and-task-speed\" class=\"anchor\" id=\"company-evaluations-report-improvements-in-success-rates-and-task-speed\"\u003e\u003c/a\u003eCompany evaluations report improvements in success rates and task speed\u003c/h2\u003e\n\u003cp\u003eThe published performance figures are results reported by Google. They include proof-of-concept results from an automotive production line. The company also announced its own evaluations against the previous baseline model.\u003c/p\u003e\n\u003cdiv class=\"overflow-x-auto\"\u003e\u003ctable\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth\u003eEvaluation metric\u003c/th\u003e\n\u003cth\u003eCompany-reported figure\u003c/th\u003e\n\u003cth\u003eScope and conditions\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Evaluation metric\"\u003eSuccess rate for picking and packing automotive parts\u003c/td\u003e\n\u003ctd data-label=\"Company-reported figure\"\u003e96%\u003c/td\u003e\n\u003ctd data-label=\"Scope and conditions\"\u003eApptronik Apollo 2 proof of concept on a Mercedes-Benz production line\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Evaluation metric\"\u003eImprovement in task success rate\u003c/td\u003e\n\u003ctd data-label=\"Company-reported figure\"\u003e30~100% higher than the previous baseline model\u003c/td\u003e\n\u003ctd data-label=\"Scope and conditions\"\u003eGoogle's own evaluation, with differences by task type\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd data-label=\"Evaluation metric\"\u003eImprovement in task completion speed\u003c/td\u003e\n\u003ctd data-label=\"Company-reported figure\"\u003e20~30% faster than the previous baseline model\u003c/td\u003e\n\u003ctd data-label=\"Scope and conditions\"\u003eResults against the previous baseline model in Google's own evaluation\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\u003c/div\u003e\n\u003cp\u003eThese figures appeared in the November 10, 2026 announcement. The 96% success rate applies to picking and packing automotive parts. It is not a success rate that applies to all robots and all tasks.\u003c/p\u003e\n\u003cp\u003eThe improvement in success rate is a relative figure measured against a comparison baseline. It should not be read as an increase of 30~100 percentage points. The improvement in task completion speed must also be distinguished from a reduction in time taken. \u003ca href=\"https://deepmind.google/blog/gemini-robotics-2-5-accelerates-real-world-physical-intelligence/\"\u003eGoogle DeepMind announcement\u003c/a\u003e\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#migration-procedures-and-independent-performance-measurements-not-confirmed\" class=\"anchor\" id=\"migration-procedures-and-independent-performance-measurements-not-confirmed\"\u003e\u003c/a\u003eMigration procedures and independent performance measurements not confirmed\u003c/h2\u003e\n\u003cp\u003eOfficial sources did not specify migration procedures for users of existing models. Whether existing access permissions allow use of the new model has not been confirmed either. Anyone preparing to adopt it should check the migration requirements separately.\u003c/p\u003e\n\u003cp\u003eNo independent performance measurements from Artificial Analysis were presented either. The company's own evaluations must be distinguished from independent measurements. There is no basis for labeling the figures currently presented as results verified by an independent organization.\u003c/p\u003e\n\u003ch2\u003e\n\u003ca href=\"#terms-to-check-before-adoption-in-korea\" class=\"anchor\" id=\"terms-to-check-before-adoption-in-korea\"\u003e\u003c/a\u003eTerms to check before adoption in Korea\u003c/h2\u003e\n\u003cp\u003eAnyone considering adoption in Korea should first check whether it is available there. Access for partners must be distinguished from a public API for general developers. The following items should be checked before assessing use or costs.\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eWhether it is available in Korea and whether you qualify for partner access\u003c/li\u003e\n\u003cli\u003eThe release timeline for a public API for general developers and the exact model ID\u003c/li\u003e\n\u003cli\u003eApplicable pricing and usage limits\u003c/li\u003e\n\u003cli\u003eProcedures required to migrate from an existing model\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eThe sources to check are Google DeepMind's official announcement and model information page. Terms that are not currently public must be checked in subsequent official updates. When reviewing performance, the task conditions in the company's evaluations also need to be checked.\u003c/p\u003e\n","tags":["AI","AI Development","Robots"],"faqs":[{"question":"Can regular developers in Korea use Gemini Robotics 2.5 right away?","answer":"Availability in Korea has not been disclosed. Google's robotics partners are the currently stated target users. The timeline for a public API for regular developers and the model ID have not been confirmed either."},{"question":"What are the pricing and usage limits for use in Korea?","answer":"Pricing and usage limits have not been disclosed. Availability in Korea has not been specified either, so there is no basis for calculating the cost of using it in Korea."},{"question":"Can the 96% success rate be used as the expected performance for robot deployments in Korea?","answer":"The 96% figure is the task success rate from a specific proof of concept reported by Google. It comes from Apollo 2 picking and packing automotive parts on a Mercedes-Benz production line. There is no evidence that other robots or tasks in Korea would achieve the same success rate."}],"sources":[{"url":"https://deepmind.google/blog/gemini-robotics-2-5-accelerates-real-world-physical-intelligence/","title":"Gemini Robotics 2.5 accelerates real-world general dexterity - Google DeepMind","type":"source"},{"url":"https://blog.google/intl/ko-kr/company-news/technology/gemini-robotics-2-5-kr/","title":"Gemini Robotics 2.5: Accelerating physical intelligence in the real world - Google","type":"source"},{"url":"https://deepmind.google/models/gemini-robotics/","title":"Gemini Robotics - Google DeepMind - Google DeepMind","type":"source"},{"url":"https://deepmind.google/models/gemini-robotics/vla/","title":"Gemini Robotics 2.5 - Google DeepMind - Google DeepMind","type":"source"}],"images":[{"id":2735,"url":"https://injoys.com/rails/active_storage/blobs/proxy/eyJfcmFpbHMiOnsiZGF0YSI6NDE5NjQsInB1ciI6ImJsb2JfaWQifX0=--e9e967cc7cc1d2a0d14c6ce00bc7cc5d986a6466/ai-e469df9b.webp","is_representative":true,"generation_method":"ai_photo","license":"ai_generated","mime_type":"image/webp","width":1536,"height":1024,"translations":{"ko":{"alt":"작업자가 지켜보는 가운데 휴머노이드 로봇이 집게로 금속 부품을 들어 완충재가 든 상자 위로 옮기고 있다.","caption":"Google DeepMind는 Gemini Robotics 2.5를 로보틱스 파트너사에 순차 제공한다.","description":null},"en":{"alt":"A humanoid robot lifts a metal part over a foam-lined crate with its gripper as a worker watches.","caption":"Google DeepMind is rolling out Gemini Robotics 2.5 to robotics partners.","description":null},"ja":{"alt":"作業員が見守る中、ヒューマノイドロボットがグリッパーで金属部品を持ち上げ、緩衝材入りの箱の上へ運んでいる。","caption":"Google DeepMindはGemini Robotics 2.5をロボティクスのパートナー企業に順次提供する。","description":null},"es":{"alt":"Un robot humanoide eleva una pieza metálica sobre una caja acolchada mientras una trabajadora observa.","caption":"Google DeepMind está dando acceso gradual a Gemini Robotics 2.5 a sus socios de robótica.","description":null},"id":{"alt":"Robot humanoid mengangkat komponen logam di atas peti berlapis busa, disaksikan seorang pekerja.","caption":"Google DeepMind memberikan akses Gemini Robotics 2.5 secara bertahap kepada mitra robotika.","description":null},"pt":{"alt":"Um robô humanoide ergue uma peça metálica sobre uma caixa com espuma enquanto uma trabalhadora observa.","caption":"O Google DeepMind está disponibilizando o Gemini Robotics 2.5 gradualmente a parceiros de robótica.","description":null},"zh-hant":{"alt":"一名工作人員在旁觀察，人形機器人用夾爪提起金屬零件，移至鋪有泡棉的箱子上方。","caption":"Google DeepMind正逐步向機器人技術合作夥伴提供Gemini Robotics 2.5。","description":null},"de":{"alt":"Ein humanoider Roboter hebt ein Metallteil über eine gepolsterte Kiste, während eine Mitarbeiterin zusieht.","caption":"Google DeepMind stellt Gemini Robotics 2.5 schrittweise seinen Robotikpartnern bereit.","description":null}}},{"id":2736,"url":"https://injoys.com/rails/active_storage/blobs/proxy/eyJfcmFpbHMiOnsiZGF0YSI6NDE5ODIsInB1ciI6ImJsb2JfaWQifX0=--10e1e7da5047ea465964e8fdbfc1342ef05d45fe/ai-617cb832.webp","is_representative":false,"generation_method":"ai_semi","license":"ai_generated","mime_type":"image/webp","width":1536,"height":1024,"translations":{"ko":{"alt":"작업자가 지켜보는 가운데 휴머노이드 로봇의 빈 집게가 파란 상자에 담긴 금속 부품 위로 올라가 있다.","caption":"Google은 자동차 부품 집기·포장 개념 검증에서 Apollo 2의 작업 성공률이 96%였다고 보고했다.","description":null},"en":{"alt":"A worker watches a humanoid robot’s empty gripper hover above a metal component in a blue bin.","caption":"Google reported a 96% task success rate for Apollo 2 in an automotive parts picking and packing proof of concept.","description":null},"ja":{"alt":"作業員が見守る中、ヒューマノイドロボットの空のグリッパーが、青い箱に入った金属部品の上に浮いている。","caption":"Googleは、自動車部品のピッキング・梱包の概念実証でApollo 2の作業成功率が96％だったと報告した。","description":null},"es":{"alt":"Una trabajadora observa la pinza vacía de un robot humanoide sobre una pieza metálica en una caja azul.","caption":"Google informó de un 96 % de éxito de Apollo 2 en una prueba de concepto de recogida y embalaje de piezas de automóvil.","description":null},"id":{"alt":"Seorang pekerja mengamati pencengkeram kosong robot humanoid di atas komponen logam dalam kotak biru.","caption":"Google melaporkan tingkat keberhasilan tugas Apollo 2 sebesar 96% dalam uji konsep pengambilan dan pengemasan komponen mobil.","description":null},"pt":{"alt":"Uma trabalhadora observa a garra vazia de um robô humanoide sobre uma peça metálica em uma caixa azul.","caption":"O Google relatou 96% de sucesso nas tarefas do Apollo 2 em uma prova de conceito de coleta e embalagem de peças automotivas.","description":null},"zh-hant":{"alt":"一名工作人員在旁觀察，人形機器人的空夾爪懸在藍色箱內的金屬零件上方。","caption":"Google表示，Apollo 2在汽車零件拾取與包裝的概念驗證中，任務成功率達96%。","description":null},"de":{"alt":"Eine Arbeiterin beobachtet den leeren Greifer eines humanoiden Roboters über einem Metallteil in einer blauen Kiste.","caption":"Google meldete für Apollo 2 eine Aufgabenerfolgsquote von 96 % beim Machbarkeitsnachweis zum Greifen und Verpacken von Autoteilen.","description":null}}}],"published_at":"2026-11-12T02:56:15+09:00","updated_at":"2026-11-12T02:56:15+09:00","license":"cc_by","translation_status":"reviewed","available_locales":["ko","en","ja","es"],"data_locales":["ko","en","ja","es","id","pt","zh-hant","de"],"url":"https://injoys.com/en/articles/google-gemini-robotics-2-5-partner-rollout"}