{"work":{"id":"0f06e436-0c76-4e3c-be5e-6168f6bc4336","openalex_id":null,"doi":null,"arxiv_id":"2408.01800","raw_key":null,"title":"MiniCPM-V: A GPT-4V Level MLLM on Your Phone","authors":null,"authors_text":"Yuan Yao, Tianyu Yu, Ao Zhang, Chongyi Wang, Junbo Cui, Hongji Zhu","year":2024,"venue":"cs.CV","abstract":"The recent surge of Multimodal Large Language Models (MLLMs) has fundamentally reshaped the landscape of AI research and industry, shedding light on a promising path toward the next AI milestone. However, significant challenges remain preventing MLLMs from being practical in real-world applications. The most notable challenge comes from the huge cost of running an MLLM with a massive number of parameters and extensive computation. As a result, most MLLMs need to be deployed on high-performing cloud servers, which greatly limits their application scopes such as mobile, offline, energy-sensitive, and privacy-protective scenarios. In this work, we present MiniCPM-V, a series of efficient MLLMs deployable on end-side devices. By integrating the latest MLLM techniques in architecture, pretraining and alignment, the latest MiniCPM-Llama3-V 2.5 has several notable features: (1) Strong performance, outperforming GPT-4V-1106, Gemini Pro and Claude 3 on OpenCompass, a comprehensive evaluation over 11 popular benchmarks, (2) strong OCR capability and 1.8M pixel high-resolution image perception at any aspect ratio, (3) trustworthy behavior with low hallucination rates, (4) multilingual support for 30+ languages, and (5) efficient deployment on mobile phones. More importantly, MiniCPM-V can be viewed as a representative example of a promising trend: The model sizes for achieving usable (e.g., GPT-4V) level performance are rapidly decreasing, along with the fast growth of end-side computation capacity. This jointly shows that GPT-4V level MLLMs deployed on end devices are becoming increasingly possible, unlocking a wider spectrum of real-world AI applications in the near future.","external_url":"https://arxiv.org/abs/2408.01800","cited_by_count":null,"metadata_source":"pith","metadata_fetched_at":"2026-07-10T11:37:03.161139+00:00","pith_arxiv_id":"2408.01800","created_at":"2026-05-09T22:34:07.291449+00:00","updated_at":"2026-07-10T11:37:03.161139+00:00","title_quality_ok":true,"display_title":"MiniCPM-V: A GPT-4V Level MLLM on Your Phone","render_title":"MiniCPM-V: A GPT-4V Level MLLM on Your Phone"},"hub":{"state":{"work_id":"0f06e436-0c76-4e3c-be5e-6168f6bc4336","tier":"super_hub","tier_reason":"100+ Pith inbound or 10,000+ external citations","pith_inbound_count":138,"external_cited_by_count":null,"distinct_field_count":11,"first_pith_cited_at":"2024-08-09T03:25:42+00:00","last_pith_cited_at":"2026-07-09T07:50:49+00:00","author_build_status":"needed","summary_status":"needed","contexts_status":"needed","graph_status":"needed","ask_index_status":"needed","reader_status":"not_needed","recognition_status":"not_needed","updated_at":"2026-08-21T19:19:21.316182+00:00","tier_text":"super_hub"},"tier":"super_hub","role_counts":[{"context_role":"baseline","n":16},{"context_role":"background","n":10},{"context_role":"dataset","n":2},{"context_role":"method","n":1}],"polarity_counts":[{"context_polarity":"baseline","n":16},{"context_polarity":"background","n":10},{"context_polarity":"use_dataset","n":2},{"context_polarity":"use_method","n":1}],"runs":{"ask_index":{"job_type":"ask_index","status":"succeeded","result":{"title":"MiniCPM-V: A GPT-4V Level MLLM on Your Phone","claims":[{"claim_text":"The recent surge of Multimodal Large Language Models (MLLMs) has fundamentally reshaped the landscape of AI research and industry, shedding light on a promising path toward the next AI milestone. However, significant challenges remain preventing MLLMs from being practical in real-world applications. The most notable challenge comes from the huge cost of running an MLLM with a massive number of parameters and extensive computation. As a result, most MLLMs need to be deployed on high-performing cloud servers, which greatly limits their application scopes such as mobile, offline, energy-sensitive","claim_type":"abstract","evidence_strength":"source_metadata"},{"claim_text":"This yieldsPhyAR (Physics-Anchored Reasoner). Through this targeted update, PhyAR performs physical reasoning based on visual perception without extensive structural modifications. 4 Experiments 4.1 Experimental Setup Baselines.We evaluate PhyAR against SOTA Video-LLMs, including open-source models (Qwen3.5- 35B [36], InternVL2.5-8B [9], MiniCPM-V 2.6 [40], VideoLLaMA2-7B [10], VideoLLaMA3-7B [43], Video-LLaV A-7B [21], Flash-VStream-7B [44], and Video-ChatGPT-7B [23]) and proprietary models (Ge","claim_type":"baseline","confidence":0.95,"evidence_strength":"citation_context"},{"claim_text":"MLLMs such as GPT-4V [48], LLaVa [35, 36], Qwen- VL [65], InternVL [7] and MiniCPM-V [71] have shown promising advancements in multimodal information under- standing and reasoning. Typically, an MLLM consists of an LLM, a vision encoder, and a projector that bridges the two components by transforming raw multimodal inputs into vectors compatible with the LLM [72]. Multimodal Retrieval Early multimodal retrieval tasks focused on single-modal [73] or cross-modal retrieval [61]. Recently, the expan","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"MiniCPM-o 4.5 advances this trajectory toward more human-like full-duplex interaction by enabling simultaneous perception and response. 1 Introduction Progress in multimodal large language models (MLLMs) has enabled increasingly rich interaction over images, speech, video, and text, bringing AI systems closer to more natural forms of com- munication [1, 2, 3, 4] (Figure 2). The main challenge towards human-like interaction now is no longer modality coverage or response latency alone, but the und","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"(w/o sub) TempCompass MVBench WildVideo LVBench LongVideoBench Proprietary LMMs GPT-4o [15] 34 61.2 67.9 - 57.5 62.1 30.8 66.7 Open-Source LMMs LLaV A-OneVision-7B [19] 32.4 33.8 58.2 - 56.7 - - - ShareGPT4Video-8B [5] - - - - - - - 39.7 LLaV A-Video-7B [56] 35.7 36.1 63.7 65.5 62.1 53.4 - 59.5 VILA-1.5-8B [24] 28.9 20.9 - 58.8 - - - - VideoLLaMA 3-7B [51] - 46.0 61.0 - - - 45.3 - MiniCPM-V 2.6-8B [47] - - 59.7 59.6 44.7 46.4 43.5 - mPLUG-Owl3-8B [48] - - 53.5 - 54.5 - - 59.8 InternVL2.5-8B [6] ","claim_type":"baseline","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"Building on the success of LLMs [1, 12, 38, 52, 56], Large Multimodal Models (LMMs) have advanced visual lan- guage understanding rapidly [24]. LLaV A [32-34] popu- larized instruction tuning on GPT-4 [1] curated data and es- tablished a widely adopted recipe for constructing LMMs, inspiring many follow up systems [6, 7, 63]. Fueled by short-video applications and video generation, video un- derstanding [55, 59, 62] has emerged as a focal area, yet spatio-temporal joint modeling remains challeng","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"66 19.78 29.49 39.41 13.71 29.37 Open-sourceQwen3-VL-8B-Instruct [2] 76824.27 14.31 33.26 23.08 44.47 27.45Qwen3-VL-4B-Instruct 76823.26 13.14 31.11 22.92 34.17 25.15VideoLLaMA-7B [50] 76820.06 10.98 16.92 14.00 28.54 17.96InternVideo2.5-8B [4] 51218.12 10.39 19.83 9.85 26.21 16.68Video-R1-7B [8] 76819.26 8.04 23.25 18.62 33.01 20.40MiniCPM-V4.5-8B [47] 51216.38 10.14 13.50 16.18 27.27 16.48GLM-4.6v-Flash-9B [30] 5122.27 1.18 3.76 3.54 13.20 4.62 5.1 Experimental setting Implementation detailsWe","claim_type":"baseline","confidence":0.9,"evidence_strength":"citation_context"}],"why_cited":"Pith tracks MiniCPM-V: A GPT-4V Level MLLM on Your Phone because it crossed a citation-hub threshold. Current citing contexts most often use it as baseline evidence (15 contexts).","role_counts":[{"n":15,"context_role":"baseline"},{"n":10,"context_role":"background"},{"n":2,"context_role":"dataset"},{"n":1,"context_role":"method"}]},"error":null,"updated_at":"2026-05-23T07:24:18.059084+00:00"},"author_expand":{"job_type":"author_expand","status":"succeeded","result":{"authors_linked":[{"id":"c5cad999-5dd6-4cd3-b762-7d388e94ab92","orcid":null,"display_name":"Yuan Yao"},{"id":"54c37964-6371-4a32-9b4e-2bc44f28d98e","orcid":null,"display_name":"Tianyu Yu"},{"id":"3522b2d4-aeda-43dd-a4b2-86c2bad175d2","orcid":null,"display_name":"Ao Zhang"},{"id":"09c7d292-7169-4320-8a4f-03958c9d7930","orcid":null,"display_name":"Chongyi Wang"},{"id":"f44fef0c-d103-46cc-a65b-1d76899b0ff5","orcid":null,"display_name":"Junbo Cui"},{"id":"e3d1b679-5a8a-4103-902e-dcfeff19edd2","orcid":null,"display_name":"Hongji Zhu"}]},"error":null,"updated_at":"2026-05-23T07:24:18.051834+00:00"},"context_extract":{"job_type":"context_extract","status":"succeeded","result":{"enqueued_papers":25},"error":null,"updated_at":"2026-05-14T13:41:14.532430+00:00"},"graph_features":{"job_type":"graph_features","status":"succeeded","result":{"co_cited":[{"title":"Qwen2.5-VL Technical Report","work_id":"69dffacb-bfe8-442d-be86-48624c60426f","shared_citers":28},{"title":"Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution","work_id":"8abcfe4f-e0fb-44b7-9123-448fac95f90a","shared_citers":24},{"title":"LLaVA-OneVision: Easy Visual Task Transfer","work_id":"f5f2452b-f2a9-49ac-b38d-c76e18cdfe49","shared_citers":19},{"title":"InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models","work_id":"fe8637aa-12bc-4434-8d36-9f57b5eebcbe","shared_citers":18},{"title":"Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling","work_id":"ee70bdc8-4656-4849-ada7-ce42a2278d70","shared_citers":12},{"title":"Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context","work_id":"80e3e977-f1bb-4c83-8d0c-1ab0a0c5c3f1","shared_citers":12},{"title":"Qwen3-VL Technical Report","work_id":"1fe243aa-e3c0-4da6-b391-4cbcfc88d5c0","shared_citers":12},{"title":"Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond","work_id":"cbc2bb21-b6bb-46c0-80bf-107e195ffe10","shared_citers":12},{"title":"GPT-4o System Card","work_id":"f37bf1c7-4964-4e56-9762-d20da8d9009f","shared_citers":11},{"title":"GPT-4 Technical Report","work_id":"b928e041-6991-4c08-8c81-0359e4097c7b","shared_citers":11},{"title":"InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency","work_id":"b8f5e260-fff5-444e-bcf5-2c42cfefd83d","shared_citers":11},{"title":"VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understanding","work_id":"38f52461-37fd-4266-bc46-9dea31be2824","shared_citers":11},{"title":"Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities","work_id":"008df105-2fdd-45d8-857a-8e35868aecb6","shared_citers":10},{"title":"LLaVA-Video: Video Instruction Tuning With Synthetic Data","work_id":"e598f516-d992-449a-ab6d-6c788b3a1d7b","shared_citers":10},{"title":"VideoLLaMA 2: Advancing Spatial-Temporal Modeling and Audio Understanding in Video-LLMs","work_id":"ccfc3f89-c510-45f1-8a35-ed1a56c0ae5c","shared_citers":10},{"title":"Gemini: A Family of Highly Capable Multimodal Models","work_id":"83f7c85b-3f11-450f-ac0c-64d9745220b2","shared_citers":8},{"title":"Qwen3 Technical Report","work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e","shared_citers":8},{"title":"The Llama 3 Herd of Models","work_id":"1549a635-88af-4ac1-acfe-51ae7bb53345","shared_citers":8},{"title":"DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models","work_id":"c5006563-f3ec-438a-9e35-b7b484f34828","shared_citers":7},{"title":"MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models","work_id":"806d2e73-71b3-4d56-87e0-39d571cc15d6","shared_citers":7},{"title":"DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning","work_id":"e6b75ad5-2877-4168-97c8-710407094d20","shared_citers":6},{"title":"GLM-4.5V and GLM-4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning","work_id":"366607ba-e4ea-4726-98c3-63356e32351c","shared_citers":6},{"title":"Long Context Transfer from Language to Vision","work_id":"52f1b946-568f-4819-9d8a-a87296f8852d","shared_citers":6},{"title":"Mantis: Interleaved multi-image instruction tuning","work_id":"ca7dd196-dc3e-4e3c-af1e-c0b02fc0efed","shared_citers":6}],"time_series":[{"n":3,"year":2024},{"n":4,"year":2025},{"n":45,"year":2026}],"dependency_candidates":[]},"error":null,"updated_at":"2026-05-14T13:41:08.855883+00:00"},"identity_refresh":{"job_type":"identity_refresh","status":"succeeded","result":{"items":[{"title":"Qwen3 Technical Report","outcome":"unchanged","work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e","resolver":"local_arxiv","confidence":0.98,"old_work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e"}],"counts":{"fixed":0,"merged":0,"unchanged":1,"quarantined":0,"needs_external_resolution":0},"errors":[],"attempted":1},"error":null,"updated_at":"2026-05-14T13:41:08.810636+00:00"},"role_polarity":{"job_type":"role_polarity","status":"succeeded","result":{"title":"MiniCPM-V: A GPT-4V Level MLLM on Your Phone","claims":[{"claim_text":"The recent surge of Multimodal Large Language Models (MLLMs) has fundamentally reshaped the landscape of AI research and industry, shedding light on a promising path toward the next AI milestone. However, significant challenges remain preventing MLLMs from being practical in real-world applications. The most notable challenge comes from the huge cost of running an MLLM with a massive number of parameters and extensive computation. As a result, most MLLMs need to be deployed on high-performing cloud servers, which greatly limits their application scopes such as mobile, offline, energy-sensitive","claim_type":"abstract","evidence_strength":"source_metadata"},{"claim_text":"This yieldsPhyAR (Physics-Anchored Reasoner). Through this targeted update, PhyAR performs physical reasoning based on visual perception without extensive structural modifications. 4 Experiments 4.1 Experimental Setup Baselines.We evaluate PhyAR against SOTA Video-LLMs, including open-source models (Qwen3.5- 35B [36], InternVL2.5-8B [9], MiniCPM-V 2.6 [40], VideoLLaMA2-7B [10], VideoLLaMA3-7B [43], Video-LLaV A-7B [21], Flash-VStream-7B [44], and Video-ChatGPT-7B [23]) and proprietary models (Ge","claim_type":"baseline","confidence":0.95,"evidence_strength":"citation_context"},{"claim_text":"MLLMs such as GPT-4V [48], LLaVa [35, 36], Qwen- VL [65], InternVL [7] and MiniCPM-V [71] have shown promising advancements in multimodal information under- standing and reasoning. Typically, an MLLM consists of an LLM, a vision encoder, and a projector that bridges the two components by transforming raw multimodal inputs into vectors compatible with the LLM [72]. Multimodal Retrieval Early multimodal retrieval tasks focused on single-modal [73] or cross-modal retrieval [61]. Recently, the expan","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"MiniCPM-o 4.5 advances this trajectory toward more human-like full-duplex interaction by enabling simultaneous perception and response. 1 Introduction Progress in multimodal large language models (MLLMs) has enabled increasingly rich interaction over images, speech, video, and text, bringing AI systems closer to more natural forms of com- munication [1, 2, 3, 4] (Figure 2). The main challenge towards human-like interaction now is no longer modality coverage or response latency alone, but the und","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"(w/o sub) TempCompass MVBench WildVideo LVBench LongVideoBench Proprietary LMMs GPT-4o [15] 34 61.2 67.9 - 57.5 62.1 30.8 66.7 Open-Source LMMs LLaV A-OneVision-7B [19] 32.4 33.8 58.2 - 56.7 - - - ShareGPT4Video-8B [5] - - - - - - - 39.7 LLaV A-Video-7B [56] 35.7 36.1 63.7 65.5 62.1 53.4 - 59.5 VILA-1.5-8B [24] 28.9 20.9 - 58.8 - - - - VideoLLaMA 3-7B [51] - 46.0 61.0 - - - 45.3 - MiniCPM-V 2.6-8B [47] - - 59.7 59.6 44.7 46.4 43.5 - mPLUG-Owl3-8B [48] - - 53.5 - 54.5 - - 59.8 InternVL2.5-8B [6] ","claim_type":"baseline","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"Building on the success of LLMs [1, 12, 38, 52, 56], Large Multimodal Models (LMMs) have advanced visual lan- guage understanding rapidly [24]. LLaV A [32-34] popu- larized instruction tuning on GPT-4 [1] curated data and es- tablished a widely adopted recipe for constructing LMMs, inspiring many follow up systems [6, 7, 63]. Fueled by short-video applications and video generation, video un- derstanding [55, 59, 62] has emerged as a focal area, yet spatio-temporal joint modeling remains challeng","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"66 19.78 29.49 39.41 13.71 29.37 Open-sourceQwen3-VL-8B-Instruct [2] 76824.27 14.31 33.26 23.08 44.47 27.45Qwen3-VL-4B-Instruct 76823.26 13.14 31.11 22.92 34.17 25.15VideoLLaMA-7B [50] 76820.06 10.98 16.92 14.00 28.54 17.96InternVideo2.5-8B [4] 51218.12 10.39 19.83 9.85 26.21 16.68Video-R1-7B [8] 76819.26 8.04 23.25 18.62 33.01 20.40MiniCPM-V4.5-8B [47] 51216.38 10.14 13.50 16.18 27.27 16.48GLM-4.6v-Flash-9B [30] 5122.27 1.18 3.76 3.54 13.20 4.62 5.1 Experimental setting Implementation detailsWe","claim_type":"baseline","confidence":0.9,"evidence_strength":"citation_context"}],"why_cited":"Pith tracks MiniCPM-V: A GPT-4V Level MLLM on Your Phone because it crossed a citation-hub threshold. Current citing contexts most often use it as baseline evidence (15 contexts).","role_counts":[{"n":15,"context_role":"baseline"},{"n":10,"context_role":"background"},{"n":2,"context_role":"dataset"},{"n":1,"context_role":"method"}]},"error":null,"updated_at":"2026-05-23T07:24:18.055821+00:00"},"summary_claims":{"job_type":"summary_claims","status":"succeeded","result":{"title":"MiniCPM-V: A GPT-4V Level MLLM on Your Phone","claims":[{"claim_text":"The recent surge of Multimodal Large Language Models (MLLMs) has fundamentally reshaped the landscape of AI research and industry, shedding light on a promising path toward the next AI milestone. However, significant challenges remain preventing MLLMs from being practical in real-world applications. The most notable challenge comes from the huge cost of running an MLLM with a massive number of parameters and extensive computation. As a result, most MLLMs need to be deployed on high-performing cloud servers, which greatly limits their application scopes such as mobile, offline, energy-sensitive","claim_type":"abstract","evidence_strength":"source_metadata"}],"why_cited":"Pith tracks MiniCPM-V: A GPT-4V Level MLLM on Your Phone because it crossed a citation-hub threshold.","role_counts":[]},"error":null,"updated_at":"2026-05-14T13:41:19.003325+00:00"}},"summary":{"title":"MiniCPM-V: A GPT-4V Level MLLM on Your Phone","claims":[{"claim_text":"The recent surge of Multimodal Large Language Models (MLLMs) has fundamentally reshaped the landscape of AI research and industry, shedding light on a promising path toward the next AI milestone. However, significant challenges remain preventing MLLMs from being practical in real-world applications. The most notable challenge comes from the huge cost of running an MLLM with a massive number of parameters and extensive computation. As a result, most MLLMs need to be deployed on high-performing cloud servers, which greatly limits their application scopes such as mobile, offline, energy-sensitive","claim_type":"abstract","evidence_strength":"source_metadata"}],"why_cited":"Pith tracks MiniCPM-V: A GPT-4V Level MLLM on Your Phone because it crossed a citation-hub threshold.","role_counts":[]},"graph":{"co_cited":[{"title":"Qwen2.5-VL Technical Report","work_id":"69dffacb-bfe8-442d-be86-48624c60426f","shared_citers":28},{"title":"Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution","work_id":"8abcfe4f-e0fb-44b7-9123-448fac95f90a","shared_citers":24},{"title":"LLaVA-OneVision: Easy Visual Task Transfer","work_id":"f5f2452b-f2a9-49ac-b38d-c76e18cdfe49","shared_citers":19},{"title":"InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models","work_id":"fe8637aa-12bc-4434-8d36-9f57b5eebcbe","shared_citers":18},{"title":"Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling","work_id":"ee70bdc8-4656-4849-ada7-ce42a2278d70","shared_citers":12},{"title":"Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context","work_id":"80e3e977-f1bb-4c83-8d0c-1ab0a0c5c3f1","shared_citers":12},{"title":"Qwen3-VL Technical Report","work_id":"1fe243aa-e3c0-4da6-b391-4cbcfc88d5c0","shared_citers":12},{"title":"Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond","work_id":"cbc2bb21-b6bb-46c0-80bf-107e195ffe10","shared_citers":12},{"title":"GPT-4o System Card","work_id":"f37bf1c7-4964-4e56-9762-d20da8d9009f","shared_citers":11},{"title":"GPT-4 Technical Report","work_id":"b928e041-6991-4c08-8c81-0359e4097c7b","shared_citers":11},{"title":"InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency","work_id":"b8f5e260-fff5-444e-bcf5-2c42cfefd83d","shared_citers":11},{"title":"VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understanding","work_id":"38f52461-37fd-4266-bc46-9dea31be2824","shared_citers":11},{"title":"Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities","work_id":"008df105-2fdd-45d8-857a-8e35868aecb6","shared_citers":10},{"title":"LLaVA-Video: Video Instruction Tuning With Synthetic Data","work_id":"e598f516-d992-449a-ab6d-6c788b3a1d7b","shared_citers":10},{"title":"VideoLLaMA 2: Advancing Spatial-Temporal Modeling and Audio Understanding in Video-LLMs","work_id":"ccfc3f89-c510-45f1-8a35-ed1a56c0ae5c","shared_citers":10},{"title":"Gemini: A Family of Highly Capable Multimodal Models","work_id":"83f7c85b-3f11-450f-ac0c-64d9745220b2","shared_citers":8},{"title":"Qwen3 Technical Report","work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e","shared_citers":8},{"title":"The Llama 3 Herd of Models","work_id":"1549a635-88af-4ac1-acfe-51ae7bb53345","shared_citers":8},{"title":"DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models","work_id":"c5006563-f3ec-438a-9e35-b7b484f34828","shared_citers":7},{"title":"MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models","work_id":"806d2e73-71b3-4d56-87e0-39d571cc15d6","shared_citers":7},{"title":"DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning","work_id":"e6b75ad5-2877-4168-97c8-710407094d20","shared_citers":6},{"title":"GLM-4.5V and GLM-4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning","work_id":"366607ba-e4ea-4726-98c3-63356e32351c","shared_citers":6},{"title":"Long Context Transfer from Language to Vision","work_id":"52f1b946-568f-4819-9d8a-a87296f8852d","shared_citers":6},{"title":"Mantis: Interleaved multi-image instruction tuning","work_id":"ca7dd196-dc3e-4e3c-af1e-c0b02fc0efed","shared_citers":6}],"time_series":[{"n":3,"year":2024},{"n":4,"year":2025},{"n":45,"year":2026}],"dependency_candidates":[]},"authors":[{"id":"3522b2d4-aeda-43dd-a4b2-86c2bad175d2","orcid":null,"display_name":"Ao Zhang","source":"manual","import_confidence":0.72},{"id":"09c7d292-7169-4320-8a4f-03958c9d7930","orcid":null,"display_name":"Chongyi Wang","source":"manual","import_confidence":0.72},{"id":"e3d1b679-5a8a-4103-902e-dcfeff19edd2","orcid":null,"display_name":"Hongji Zhu","source":"manual","import_confidence":0.72},{"id":"f44fef0c-d103-46cc-a65b-1d76899b0ff5","orcid":null,"display_name":"Junbo Cui","source":"manual","import_confidence":0.72},{"id":"54c37964-6371-4a32-9b4e-2bc44f28d98e","orcid":null,"display_name":"Tianyu Yu","source":"manual","import_confidence":0.72},{"id":"c5cad999-5dd6-4cd3-b762-7d388e94ab92","orcid":null,"display_name":"Yuan Yao","source":"manual","import_confidence":0.72}]}}