# open models insights Rangoon Insights reporting and guides tagged open models. Canonical: https://rangoon.ai/insights/tags/open-models/ ## Topics - [Agent systems](https://rangoon.ai/insights/topics/agents/): Agent runtimes, handoffs, tools, and the boundaries that make them reviewable. - [Governance](https://rangoon.ai/insights/topics/governance/): Policy, identity, authorization, approvals, and evidence for consequential work. - [Infrastructure](https://rangoon.ai/insights/topics/infrastructure/): Model serving, sandboxes, tool services, and deployment boundaries. - [Models](https://rangoon.ai/insights/topics/models/): Inference choices, local serving, compatibility, and operational tradeoffs. ## Stories - [NASA and IBM launched a lunar foundation model with code and model access for science work](https://rangoon.ai/insights/archive-nasa-ibm-lunar-foundation-model-september-2026/) NASA and IBM launched a lunar science foundation model on September 10, 2026, trained primarily on Lunar Reconnaissance Orbiter data and released with code. - [NVIDIA introduced an Earth-2 open-model family for weather AI tools and workflows](https://rangoon.ai/insights/archive-nvidia-earth2-open-weather-models-january-2026/) NVIDIA announced an open Earth-2 model family on January 26, 2026, covering accelerated weather AI tools and models for medium-range forecasting and nowcasting. - [Ultralytics released YOLO26 with optional NMS-free real-time vision inference](https://rangoon.ai/insights/archive-ultralytics-yolo26-vision-release-january-2026/) Ultralytics released YOLO26 on January 14, 2026, adding a unified vision family with an optional end-to-end detection route that omits NMS. - [Ai2 released Molmo 2 with open video-grounding models, data and checkpoints](https://rangoon.ai/insights/archive-ai2-molmo2-video-grounding-december-2025/) Ai2 released Molmo 2 on December 11, 2025, adding 4B, 8B and fully open 7B video-language variants with pointing and tracking artifacts. - [Ai2 introduced the OlmoEarth platform for open Earth-observation models and planetary-analysis tooling](https://rangoon.ai/insights/archive-ai2-olmoearth-platform-november-2025/) Ai2 introduced the OlmoEarth platform on November 4, 2025, pairing open Earth-observation model work with tools intended for planetary insights. - [IBM released Granite-Docling-258M as a compact document-conversion VLM](https://rangoon.ai/insights/archive-ibm-granite-docling-258m-september-2025/) IBM released Granite-Docling-258M on September 24, 2025, an Apache-2.0 258M vision-language model for structured document conversion. - [Tencent published the Hunyuan-MT technical report for multilingual translation models](https://rangoon.ai/insights/archive-tencent-hunyuan-mt-technical-report-september-2025/) Tencent published the Hunyuan-MT technical report on September 5, 2025, describing 33-language translation and a separate Chimera refinement model. - [IBM and NASA opened Surya for heliophysics research, with a NASA technical record of its solar inputs](https://rangoon.ai/insights/archive-ibm-nasa-surya-heliophysics-august-2025/) IBM announced Surya on August 20, 2025 as an open heliophysics foundation model developed with NASA for work with solar-observation data. - [Mistral released Voxtral speech-understanding models in open-source and production sizes](https://rangoon.ai/insights/archive-mistral-voxtral-speech-july-2025/) Mistral released Voxtral on July 15, 2025, with 24B and 3B speech-understanding models under Apache 2.0 and a separate transcription API route. - [MiniMax-M1 paired a million-token context claim with two stated thinking budgets and vendor benchmarks](https://rangoon.ai/insights/archive-minimax-m1-open-hybrid-attention-june-2025/) MiniMax introduced open-weight M1 on June 16, 2025, with a stated 1M-token context and separate 40K and 80K thinking-budget variants. - [Google released TxGemma as a Gemma 2 family for therapeutic-property research tasks](https://rangoon.ai/insights/archive-google-txgemma-therapeutics-march-2025/) Google released TxGemma on March 25, 2025: 2B, 9B and 27B Gemma 2-derived models for therapeutic-data prediction and conversational research workflows. - [AI21 positioned Jamba 1.6 for private enterprise deployment, while keeping context and benchmark claims testable](https://rangoon.ai/insights/archive-ai21-jamba-1-6-private-enterprise-march-2025/) AI21 introduced Jamba 1.6 on March 6, 2025, naming Large and Mini variants with 256K context and private on-premises or VPC deployment paths. - [Ai2 released olmOCR as an open 7B vision model for PDF extraction](https://rangoon.ai/insights/archive-ai2-olmocr-pdf-extraction-february-2025/) Ai2 released olmOCR on February 25, 2025, pairing a fine-tuned 7B vision-language model with an open PDF-to-Markdown pipeline. - [Hugging Face released SmolVLM2 as a compact family for image and video understanding](https://rangoon.ai/insights/archive-hugging-face-smolvlm2-video-february-2025/) Hugging Face released SmolVLM2 on February 20, 2025, adding 2.2B, 500M and 256M vision-language models for small-device video workflows. - [OLMo 3 made end-to-end openness a versioned supply chain](https://rangoon.ai/insights/archive-olmo-3-release-2025/) Ai2 released OLMo 3 in November 2025 with models, data, checkpoints, and training details. Complete openness still needs artifact versions and reproducible evaluation paths. - [Kimi K2 Thinking update separated a reasoning mode from a turbo API tier](https://rangoon.ai/insights/archive-moonshot-kimi-k2-thinking-november-2025/) Moonshot AI’s November 7, 2025 forum announcement named Kimi K2 Thinking; its undated project page supplies separate technical context. - [Boltz-2 put binding-affinity prediction beside structure prediction](https://rangoon.ai/insights/archive-boltz2-binding-affinity-2025/) The Boltz team introduced Boltz-2 on June 6, 2025, reporting an open model for joint biomolecular structure and binding-affinity prediction. - [Gemma 3n framed on-device multimodality as an artifact-provenance problem](https://rangoon.ai/insights/archive-google-gemma-3n-may-2025/) Google introduced Gemma 3n on May 20, 2025, describing an open multimodal model designed for phones, laptops, and tablets. - [Isaac GR00T N1 made humanoid action generation a deployment-boundary question](https://rangoon.ai/insights/archive-isaac-groot-n1-humanoid-2025/) NVIDIA introduced Isaac GR00T N1 on March 18, 2025 as an open humanoid-robot foundation model with a reported dual-system architecture. - [OLMo 2 32B made fully open scaling a training-budget record](https://rangoon.ai/insights/archive-olmo-2-32b-2025/) Ai2 released OLMo 2 32B on March 13, 2025 and described open training artifacts plus reported comparisons. A scale claim needs compute, data, checkpoint, and scorer context. ## More information - [Documentation index](https://rangoon.ai/llms.txt) - [AI access and policies](https://rangoon.ai/ai/) - [Source repository](https://github.com/hypler-dev/rangoon)