# OpenModelsHQ > Make sense of the AI ecosystem. ## Data - [All models as JSON](https://openmodelshq.com/data/models.json): 562 open-weight models with size, license, memory estimate and cheapest API price - [Hardware as JSON](https://openmodelshq.com/data/hardware.json): 53 GPUs and devices ## Sections - [About OpenModelsHQ](https://openmodelshq.com/about/): the publication, scope and roadmap - [Latest](https://openmodelshq.com/latest/): editorial picks and recent model records - [Research](https://openmodelshq.com/research/): evidence, guides and infrastructure economics - [Infrastructure](https://openmodelshq.com/infrastructure/) - [Ecosystem](https://openmodelshq.com/ecosystem/): labs, countries and listed companies - [Training data](https://openmodelshq.com/training-data/): source-based provider directory - [Evaluations](https://openmodelshq.com/evaluations/): local output recorder and evaluation frameworks - [Startup programmes](https://openmodelshq.com/perks/): public programmes and eligibility sources - [Contributors](https://openmodelshq.com/contributors/): opt-in recognition for accepted contributions - [Jobs](https://openmodelshq.com/jobs/): listings from official employer feeds - [Suggest a source](https://openmodelshq.com/submit/): compose an email suggestion - [Models](https://openmodelshq.com/models/) - [Hardware](https://openmodelshq.com/hardware/) - [Labs](https://openmodelshq.com/labs/) - [API pricing](https://openmodelshq.com/pricing/): 71 open models with a hosted API, every provider, USD and INR per million tokens - [Releases](https://openmodelshq.com/releases/): every model, newest first - [What shipped in the last 30 days](https://openmodelshq.com/new/): 20 models from the last 30 days of releases, grouped by week - [Small models](https://openmodelshq.com/models/small/): 291 models under 10B parameters, most downloaded first, with 8 GB and 16 GB fit - [Inference providers](https://openmodelshq.com/providers/): 64 providers that serve open models, with models served, median price and precision - [Methodology](https://openmodelshq.com/about/methodology/): how memory, fit and prices are calculated ## GPU useful life and listed companies - [GPU interconnects](https://openmodelshq.com/compute/gpu-interconnects/): NVLink domains, directional bandwidth and cloud topology checklist - [GPU useful life](https://openmodelshq.com/hardware/useful-life/): launch dates, scoped evidence and an ownership cost calculator - [Operator survey](https://openmodelshq.com/hardware/useful-life/survey/): voluntary reports, reviewed before publication - [Listed infrastructure companies](https://openmodelshq.com/companies/): selected listings and source filings across the AI supply chain ## Agents and harnesses - [Agents](https://openmodelshq.com/agents/): source-based product and software profiles - [Agent harnesses](https://openmodelshq.com/agents/harnesses/): tools, context and execution around a model - [Manus](https://openmodelshq.com/agents/manus/): Hosted agent product. A hosted agent for tasks involving research, browser work, files and finished deliverables. - [Deep Agents](https://openmodelshq.com/agents/deep-agents/): Agent harness. LangChain's harness for agents that need planning, tools, context management and delegated work. - [OpenHands](https://openmodelshq.com/agents/openhands/): Software-agent platform and SDK. An ecosystem of tools for building and running agents that work with software and code. ## Services - [Deployment consulting](https://openmodelshq.com/services/consulting/): model choice, evaluation and deployment planning - [GPU sourcing](https://openmodelshq.com/services/gpu-brokering/): rental or purchase requirements and supplier discussions ## API pricing - [Open model API pricing](https://openmodelshq.com/pricing/) - [Alibaba Qwen API pricing](https://openmodelshq.com/pricing/qwen/): 6 models, lowest blended $0.0465 per 1M tokens - [ByteDance Seed API pricing](https://openmodelshq.com/pricing/bytedance-seed/): 1 model, lowest blended $0.125 per 1M tokens - [Cohere Labs API pricing](https://openmodelshq.com/pricing/cohere/): 2 models, lowest blended $0.0656 per 1M tokens - [DeepSeek API pricing](https://openmodelshq.com/pricing/deepseek/): 10 models, lowest blended $0.0314 per 1M tokens - [Google API pricing](https://openmodelshq.com/pricing/google/): 5 models, lowest blended $0.0208 per 1M tokens - [IBM Granite API pricing](https://openmodelshq.com/pricing/ibm/): 1 model, lowest blended $0.0825 per 1M tokens - [Meta API pricing](https://openmodelshq.com/pricing/meta/): 5 models, lowest blended $0.0705 per 1M tokens - [Microsoft API pricing](https://openmodelshq.com/pricing/microsoft/): 1 model, lowest blended $0.07 per 1M tokens - [MiniMax API pricing](https://openmodelshq.com/pricing/minimax/): 4 models, lowest blended $0.1312 per 1M tokens - [Mistral AI API pricing](https://openmodelshq.com/pricing/mistral/): 1 model, lowest blended $1.05 per 1M tokens - [Moonshot AI API pricing](https://openmodelshq.com/pricing/moonshot/): 8 models, lowest blended $0.7021 per 1M tokens - [NVIDIA API pricing](https://openmodelshq.com/pricing/nvidia/): 4 models, lowest blended $0.0488 per 1M tokens - [OpenAI API pricing](https://openmodelshq.com/pricing/openai/): 3 models, lowest blended $0.0225 per 1M tokens - [StepFun API pricing](https://openmodelshq.com/pricing/stepfun/): 2 models, lowest blended $0.15 per 1M tokens - [Tencent Hunyuan API pricing](https://openmodelshq.com/pricing/tencent/): 4 models, lowest blended $0.231 per 1M tokens - [Xiaomi MiMo API pricing](https://openmodelshq.com/pricing/xiaomi/): 4 models, lowest blended $0.14 per 1M tokens - [Z.ai (Zhipu) API pricing](https://openmodelshq.com/pricing/zai/): 10 models, lowest blended $0.0495 per 1M tokens ## Inference providers - [Inference providers](https://openmodelshq.com/providers/) - [DeepInfra](https://openmodelshq.com/providers/deepinfra/): 40 models, median blended $0.2331 per 1M tokens, lowest blended price on 11 of 38 models shared with other providers, only provider for 2 - [Venice](https://openmodelshq.com/providers/venice/): 27 models, median blended $0.7713 per 1M tokens, lowest blended price on 1 of 26 models shared with other providers, only provider for 1 - [Novita](https://openmodelshq.com/providers/novita/): 26 models, median blended $0.9988 per 1M tokens, lowest blended price on 3 of 23 models shared with other providers, only provider for 3 - [Parasail](https://openmodelshq.com/providers/parasail/): 26 models, median blended $0.2762 per 1M tokens, lowest blended price on 2 of 24 models shared with other providers, only provider for 2 - [SiliconFlow](https://openmodelshq.com/providers/siliconflow/): 26 models, median blended $0.4812 per 1M tokens, lowest blended price on 2 of 24 models shared with other providers, only provider for 2 - [GMICloud](https://openmodelshq.com/providers/gmicloud/): 25 models, median blended $0.429 per 1M tokens, lowest blended price on 6 of 24 models shared with other providers, only provider for 1 - [CoreWeave](https://openmodelshq.com/providers/coreweave/): 20 models, median blended $0.2787 per 1M tokens, lowest blended price on 3 of 20 models shared with other providers - [AtlasCloud](https://openmodelshq.com/providers/atlascloud/): 18 models, median blended $0.6019 per 1M tokens, lowest blended price on 3 of 18 models shared with other providers - [DigitalOcean](https://openmodelshq.com/providers/digitalocean/): 16 models, median blended $0.6675 per 1M tokens, lowest blended price on 0 of 16 models shared with other providers - [StreamLake](https://openmodelshq.com/providers/streamlake/): 16 models, median blended $0.4401 per 1M tokens, lowest blended price on 2 of 16 models shared with other providers - [Alibaba](https://openmodelshq.com/providers/alibaba/): 13 models, median blended $1.683 per 1M tokens, lowest blended price on 0 of 12 models shared with other providers, only provider for 1 - [Cloudflare](https://openmodelshq.com/providers/cloudflare/): 13 models, median blended $0.783 per 1M tokens, lowest blended price on 1 of 12 models shared with other providers, only provider for 1 - [Nebius](https://openmodelshq.com/providers/nebius/): 13 models, median blended $0.525 per 1M tokens, lowest blended price on 0 of 12 models shared with other providers, only provider for 1 - [Baidu](https://openmodelshq.com/providers/baidu/): 12 models, median blended $0.4648 per 1M tokens, lowest blended price on 6 of 12 models shared with other providers - [Together](https://openmodelshq.com/providers/together/): 12 models, median blended $0.525 per 1M tokens, lowest blended price on 0 of 12 models shared with other providers - [BaseTen](https://openmodelshq.com/providers/baseten/): 11 models, median blended $0.525 per 1M tokens, lowest blended price on 0 of 11 models shared with other providers - [Relace](https://openmodelshq.com/providers/relace/): 11 models, median blended $1.05 per 1M tokens, lowest blended price on 3 of 11 models shared with other providers - [Wafer](https://openmodelshq.com/providers/wafer/): 11 models, median blended $0.4024 per 1M tokens, lowest blended price on 2 of 11 models shared with other providers - [PrimeIntellect](https://openmodelshq.com/providers/primeintellect/): 10 models, median blended $0.0899 per 1M tokens, lowest blended price on 6 of 10 models shared with other providers - [Darkbloom](https://openmodelshq.com/providers/darkbloom/): 9 models, median blended $0.0938 per 1M tokens, lowest blended price on 2 of 9 models shared with other providers - [Fireworks](https://openmodelshq.com/providers/fireworks/): 9 models, median blended $0.3675 per 1M tokens, lowest blended price on 0 of 9 models shared with other providers - [Z.AI](https://openmodelshq.com/providers/z-ai/): 9 models, median blended $1.00 per 1M tokens, lowest blended price on 0 of 8 models shared with other providers, only provider for 1 - [AkashML](https://openmodelshq.com/providers/akashml/): 8 models, median blended $0.29 per 1M tokens, lowest blended price on 0 of 8 models shared with other providers - [Cohere](https://openmodelshq.com/providers/cohere/): 8 models, median blended $0.2375 per 1M tokens, lowest blended price on 0 of 6 models shared with other providers, only provider for 2 - [Crusoe](https://openmodelshq.com/providers/crusoe/): 8 models, median blended $0.3775 per 1M tokens, lowest blended price on 0 of 8 models shared with other providers - [DekaLLM](https://openmodelshq.com/providers/dekallm/): 8 models, median blended $0.2025 per 1M tokens, lowest blended price on 0 of 8 models shared with other providers - [Modal](https://openmodelshq.com/providers/modal/): 8 models, median blended $0.565 per 1M tokens, lowest blended price on 0 of 8 models shared with other providers - [Sail Research](https://openmodelshq.com/providers/sail-research/): 8 models, median blended $0.80 per 1M tokens, lowest blended price on 1 of 8 models shared with other providers - [InferenceNet](https://openmodelshq.com/providers/inferencenet/): 7 models, median blended $0.2625 per 1M tokens, lowest blended price on 0 of 7 models shared with other providers - [Makora](https://openmodelshq.com/providers/makora/): 7 models, median blended $0.4125 per 1M tokens, lowest blended price on 0 of 7 models shared with other providers - [Amazon Bedrock](https://openmodelshq.com/providers/amazon-bedrock/): 6 models, median blended $1.375 per 1M tokens, lowest blended price on 0 of 6 models shared with other providers - [Chutes](https://openmodelshq.com/providers/chutes/): 6 models, median blended $0.7012 per 1M tokens, lowest blended price on 0 of 6 models shared with other providers - [Inceptron](https://openmodelshq.com/providers/inceptron/): 6 models, median blended $1.4307 per 1M tokens, lowest blended price on 0 of 6 models shared with other providers - [Mancer 2](https://openmodelshq.com/providers/mancer-2/): 6 models, median blended $0.2922 per 1M tokens, lowest blended price on 0 of 6 models shared with other providers - [Morph](https://openmodelshq.com/providers/morph/): 6 models, median blended $0.5639 per 1M tokens, lowest blended price on 0 of 6 models shared with other providers - [NextBit](https://openmodelshq.com/providers/nextbit/): 6 models, median blended $0.1387 per 1M tokens, lowest blended price on 1 of 6 models shared with other providers - [Reka](https://openmodelshq.com/providers/reka/): 6 models, median blended $0.3225 per 1M tokens, lowest blended price on 0 of 6 models shared with other providers - [Friendli](https://openmodelshq.com/providers/friendli/): 5 models, median blended $1.935 per 1M tokens, lowest blended price on 0 of 5 models shared with other providers - [Google](https://openmodelshq.com/providers/google/): 5 models, median blended $0.3625 per 1M tokens, lowest blended price on 0 of 5 models shared with other providers - [Groq](https://openmodelshq.com/providers/groq/): 5 models, median blended $0.2625 per 1M tokens, lowest blended price on 0 of 4 models shared with other providers, only provider for 1 - [Mistral](https://openmodelshq.com/providers/mistral/): 5 models, median blended $1.60 per 1M tokens, lowest blended price on 0 of 4 models shared with other providers, only provider for 1 - [River](https://openmodelshq.com/providers/river/): 5 models, median blended $0.21 per 1M tokens, lowest blended price on 0 of 5 models shared with other providers - [SambaNova](https://openmodelshq.com/providers/sambanova/): 5 models, median blended $0.5725 per 1M tokens, lowest blended price on 0 of 5 models shared with other providers - [Io Net](https://openmodelshq.com/providers/io-net/): 4 models, median blended $0.2213 per 1M tokens, lowest blended price on 0 of 4 models shared with other providers - [KT](https://openmodelshq.com/providers/kt/): 4 models, median blended $0.5813 per 1M tokens, lowest blended price on 1 of 4 models shared with other providers - [Moonshot AI](https://openmodelshq.com/providers/moonshot-ai/): 4 models, median blended $2.8563 per 1M tokens, lowest blended price on 0 of 4 models shared with other providers - [Near AI](https://openmodelshq.com/providers/near-ai/): 4 models, median blended $0.2762 per 1M tokens, lowest blended price on 0 of 4 models shared with other providers - [Tenstorrent](https://openmodelshq.com/providers/tenstorrent/): 4 models, median blended $0.3594 per 1M tokens, lowest blended price on 0 of 4 models shared with other providers - [OpenInference](https://openmodelshq.com/providers/openinference/): 3 models, median blended $0.2525 per 1M tokens, lowest blended price on 0 of 3 models shared with other providers - [PrimaLabs](https://openmodelshq.com/providers/primalabs/): 3 models, median blended $5.00 per 1M tokens, lowest blended price on 0 of 3 models shared with other providers - [Azure](https://openmodelshq.com/providers/azure/): 2 models, median blended $1.3437 per 1M tokens, lowest blended price on 0 of 2 models shared with other providers - [Cerebras](https://openmodelshq.com/providers/cerebras/): 2 models, median blended $0.7825 per 1M tokens, lowest blended price on 0 of 2 models shared with other providers - [Decart](https://openmodelshq.com/providers/decart/): 2 models, median blended $1.6538 per 1M tokens, lowest blended price on 0 of 2 models shared with other providers - [DeepSeek](https://openmodelshq.com/providers/deepseek/): 2 models, median blended $0.6262 per 1M tokens, lowest blended price on 0 of 2 models shared with other providers - [Ionstream](https://openmodelshq.com/providers/ionstream/): 2 models, median blended $0.3169 per 1M tokens, lowest blended price on 0 of 2 models shared with other providers - [Minimax](https://openmodelshq.com/providers/minimax/): 2 models, median blended $0.3937 per 1M tokens, lowest blended price on 0 of 2 models shared with other providers - [Particle AI](https://openmodelshq.com/providers/particle-ai/): 2 models, median blended $0.15 per 1M tokens, lowest blended price on 0 of 2 models shared with other providers - [StepFun](https://openmodelshq.com/providers/stepfun/): 2 models, median blended $0.3106 per 1M tokens, lowest blended price on 1 of 2 models shared with other providers - [Tencent](https://openmodelshq.com/providers/tencent/): 2 models, median blended $0.7409 per 1M tokens, lowest blended price on 2 of 2 models shared with other providers - [Xiaomi](https://openmodelshq.com/providers/xiaomi/): 2 models, median blended $0.3594 per 1M tokens, lowest blended price on 0 of 2 models shared with other providers - [Ambient](https://openmodelshq.com/providers/ambient/): 1 model, median blended $0.95 per 1M tokens, lowest blended price on 0 of 1 models shared with other providers - [Doubleword](https://openmodelshq.com/providers/doubleword/): 1 model, median blended $2.15 per 1M tokens, lowest blended price on 0 of 1 models shared with other providers - [Featherless](https://openmodelshq.com/providers/featherless/): 1 model, median blended $0.7813 per 1M tokens, lowest blended price on 0 of 1 models shared with other providers - [Inferact vLLM](https://openmodelshq.com/providers/inferact-vllm/): 1 model, median blended $0.1312 per 1M tokens, lowest blended price on 1 of 1 models shared with other providers ## Countries - [Open models by country](https://openmodelshq.com/countries/) - [Canada](https://openmodelshq.com/countries/ca/): 15 tracked models from 1 labs - [China](https://openmodelshq.com/countries/cn/): 195 tracked models from 12 labs - [Finland](https://openmodelshq.com/countries/fi/): 12 tracked models from 1 labs - [France](https://openmodelshq.com/countries/fr/): 7 tracked models from 1 labs - [Germany](https://openmodelshq.com/countries/de/): 26 tracked models from 2 labs - [India](https://openmodelshq.com/countries/in/): 21 tracked models from 2 labs - [Israel](https://openmodelshq.com/countries/il/): 24 tracked models from 2 labs - [Japan](https://openmodelshq.com/countries/jp/): 15 tracked models from 1 labs - [Poland](https://openmodelshq.com/countries/pl/): 15 tracked models from 1 labs - [Singapore](https://openmodelshq.com/countries/sg/): 15 tracked models from 1 labs - [South Korea](https://openmodelshq.com/countries/kr/): 35 tracked models from 3 labs - [Spain](https://openmodelshq.com/countries/es/): 15 tracked models from 1 labs - [Switzerland](https://openmodelshq.com/countries/ch/): 15 tracked models from 1 labs - [United Arab Emirates](https://openmodelshq.com/countries/ae/): 15 tracked models from 1 labs - [United Kingdom](https://openmodelshq.com/countries/gb/): 15 tracked models from 1 labs - [United States](https://openmodelshq.com/countries/us/): 122 tracked models from 7 labs ## Model types - [Text models](https://openmodelshq.com/models/type/text/): 314 open text models. Chat and reasoning LLMs. - [Coding models](https://openmodelshq.com/models/type/coding/): 11 open coding models. Models tuned for code. - [Image models](https://openmodelshq.com/models/type/image/): 28 open image models. Text to image and editing. - [Video models](https://openmodelshq.com/models/type/video/): 35 open video models. Text and image to video. - [Audio models](https://openmodelshq.com/models/type/audio/): 54 open audio models. Speech to text and text to speech. - [Embedding models](https://openmodelshq.com/models/type/embedding/): 24 open embedding models. Search and retrieval vectors. - [Vision models](https://openmodelshq.com/models/type/vision/): 107 open vision-language models. Models that read images. ## Licenses - [Licenses](https://openmodelshq.com/licenses/) - [Apache 2.0 vs MIT: which licence is better for AI models](https://openmodelshq.com/licenses/apache-2-0-vs-mit/): Apache 2.0 vs MIT: both allow commercial use. Apache adds a patent grant, change notices and a trademark clause. A side-by-side comparison for model users. - [Apache 2.0 licence for AI models: what it allows](https://openmodelshq.com/licenses/apache-2-0/): Apache 2.0 lets you use, modify, sell and redistribute a model with notices kept and a patent grant included. What you may, must and may not do. - [Gemma Terms of Use and the Gemma 4 licence explained](https://openmodelshq.com/licenses/gemma-terms/): Gemma 1 to 3 use Google's Gemma Terms of Use with a prohibited use policy. Gemma 4 is released under plain Apache 2.0. What each version allows. - [Llama Community Licence: Llama 2, 3.x and 4 terms](https://openmodelshq.com/licenses/llama-community/): Llama licences allow commercial use with conditions: a 700 million monthly user test, naming and attribution rules, and a use policy. Versions compared. - [MIT licence for AI models: what it allows](https://openmodelshq.com/licenses/mit/): The MIT licence lets you use, modify, sell and redistribute a model if you keep the copyright notice. What it permits and what it leaves out. - [Modified MIT licences: Kimi, MiniMax and GLM thresholds](https://openmodelshq.com/licenses/modified-mit/): Kimi K2 and MiniMax M2 add a display rule to MIT above set user or revenue levels. Exact thresholds, and how K3, M2.7, M3 and GLM-5.3 differ. ## China Desk - [China Desk](https://openmodelshq.com/china/) - [Is DeepSeek safe to use? Local weights vs the app and API](https://openmodelshq.com/china/is-deepseek-safe-to-use/): Local DeepSeek weights keep your data on your machine. The app and API store it in China under DeepSeek's privacy policy. Restrictions and risks. - [Who owns DeepSeek, Moonshot AI, Zhipu (Z.ai) and MiniMax](https://openmodelshq.com/china/who-owns-deepseek-moonshot-zhipu-minimax/): High-Flyer owns DeepSeek, Alibaba took a reported 36% of Moonshot, and Zhipu and MiniMax trade in Hong Kong. Founders, funding and listings. - [Xiaomi MiMo explained: models, licences and where it fits](https://openmodelshq.com/china/xiaomi-mimo-explained/): Xiaomi's MiMo models run from a 7B reasoning model to a 1.02 trillion parameter one, mostly under MIT. Release history, API-only launches and licences. ## Run open models - [Run open models](https://openmodelshq.com/run/) - [GGUF and llama.cpp: run quantised open models](https://openmodelshq.com/run/gguf/): What the GGUF format is, how to run it with llama-cli and llama-server, how to read quant names like Q4_K_M, and how LM Studio and Ollama relate. - [Run open models with MLX on a Mac](https://openmodelshq.com/run/mlx/): How to run open models on Apple silicon with MLX and mlx-lm: install, mlx_lm.generate, the local server, mlx-community repos, unified memory and mlx-vlm. - [Run open models with Ollama](https://openmodelshq.com/run/ollama/): How to install Ollama on macOS, Linux and Windows, run open models, pick quantised tags, pull GGUF from Hugging Face and use the OpenAI-compatible API. ## Guides - [Guides](https://openmodelshq.com/guides/) - [How to choose a GPU for local LLMs](https://openmodelshq.com/guides/best-gpu-for-local-llms/): Pick a GPU for local LLMs by memory first, bandwidth second and software support third. Consumer cards, the RTX PRO 6000, Macs and the DGX Spark. - [Which open models run on the NVIDIA DGX Spark](https://openmodelshq.com/guides/dgx-spark-which-open-models-it-runs/): The DGX Spark has 128 GB of unified memory. See how that decides which open models fit, how clustering works, and when renting is the better choice. - [RTX PRO 6000 Blackwell for local AI](https://openmodelshq.com/guides/rtx-pro-6000-for-local-ai/): The RTX PRO 6000 Blackwell has 96 GB of GDDR7. Specs for each edition, what that memory makes possible, and how it compares with the RTX 5090. ## CUDA and alternatives - [CUDA and alternatives](https://openmodelshq.com/compute/) - [Metal, MPS and MLX: the Apple silicon AI stack](https://openmodelshq.com/compute/apple-metal-mlx/): How Metal, the PyTorch mps device and Apple's MLX framework fit together on a Mac, with the macOS and hardware requirements each one states. - [Huawei Ascend, CANN and other non-CUDA accelerators](https://openmodelshq.com/compute/china-ascend-cann-and-others/): What CANN is, how torch_npu and MindSpore reach Ascend NPUs, the versions vLLM Ascend pins, and where Tenstorrent, TPU and AWS Neuron stacks sit. - [CUDA explained: why it became the default for AI](https://openmodelshq.com/compute/cuda/): What CUDA is, how cuDNN, NCCL and TensorRT deepen the dependency, and what the CUDA Toolkit licence actually permits, from NVIDIA docs and EULA. - [Intel oneAPI, SYCL and PyTorch XPU support](https://openmodelshq.com/compute/intel-oneapi-sycl/): What SYCL and oneAPI are, which Intel GPUs PyTorch validates for the xpu device, and what Intel's notice says about the Intel Extension for PyTorch. - [AMD ROCm and HIP for AI: what is supported](https://openmodelshq.com/compute/rocm/): Which AMD GPUs ROCm officially supports, the PyTorch and vLLM versions AMD validates, and what HIPIFY does to CUDA source. - [Triton and Mojo: portable kernels above CUDA](https://openmodelshq.com/compute/triton-and-mojo/): OpenAI Triton as a kernel language PyTorch already depends on, and what Modular claims for Mojo and MAX, with the platform support each one documents. - [llama.cpp backends: Vulkan, CUDA, Metal, SYCL and more](https://openmodelshq.com/compute/vulkan-and-llama-cpp-backends/): The full llama.cpp backend list, which ones ship prebuilt binaries, and why Vulkan is the practical choice on consumer AMD and Intel GPUs. - [ZLUDA and CUDA translation layers: where things stand](https://openmodelshq.com/compute/zluda-and-translation-layers/): What ZLUDA translates, which AI workloads its developer reports working, and how funding changes affected its scope. ## Watch - [Watch](https://openmodelshq.com/watch/) - [Jev from TypeSafe AI: a closed model that returns typed decisions](https://openmodelshq.com/watch/jev-typesafe-ai/): Jev is a closed, API-only model from TypeSafe AI that returns typed answers with probabilities. What it accepts, costs, and which claims are untested.