Alternatives to BaseRT
Compare BaseRT alternatives for your business or organization using the curated list below. SourceForge ranks the best alternatives to BaseRT in 2026. Compare features, ratings, user reviews, pricing, and more from BaseRT competitors and alternatives in order to make an informed decision for your business.
-
1
LMCache
LMCache
LMCache is an open source Knowledge Delivery Network (KDN) designed as a caching layer for large language model serving that accelerates inference by reusing KV (key-value) caches across repeated or overlapping computations. It enables fast prompt caching, allowing LLMs to “prefill” recurring text only once and then reuse those stored KV caches, even in non-prefix positions, across multiple serving instances. This approach reduces time to first token, saves GPU cycles, and increases throughput in scenarios such as multi-round question answering or retrieval augmented generation. LMCache supports KV cache offloading (moving cache from GPU to CPU or disk), cache sharing across instances, and disaggregated prefill, which separates the prefill and decoding phases for resource efficiency. It is compatible with inference engines like vLLM and TGI and supports compressed storage, blending techniques to merge caches, and multiple backend storage options.Starting Price: Free -
2
Lucebox
Lucebox
Lucebox is a plug-and-play computer built for running local AI models and agents at full speed. Inside the custom chassis, a Ryzen AI MAX+ 395 with 128GB of unified LPDDR5X memory is paired with an RTX 3090, and the two work together through an open-source inference engine hand-tuned for exactly this hardware. The architecture is what makes it fast. Large models live in the 128GB unified memory tier, while the 3090's high-bandwidth VRAM acts as a fast tier. Speculative decoding (DFlash) and speculative prefill (PFlash) bridge the two, producing inference speeds up to 10x higher than llama.cpp on the same silicon and beating machines like the Mac Studio and DGX Spark at a fraction of their effective cost.Starting Price: $4,900 - One time payment -
3
EXAONE Deep
LG
EXAONE Deep is a series of reasoning-enhanced language models developed by LG AI Research, featuring parameter sizes of 2.4 billion, 7.8 billion, and 32 billion. These models demonstrate superior capabilities in various reasoning tasks, including math and coding benchmarks. Notably, EXAONE Deep 2.4B outperforms other models of comparable size, EXAONE Deep 7.8B surpasses both open-weight models of similar scale and the proprietary reasoning model OpenAI o1-mini, and EXAONE Deep 32B shows competitive performance against leading open-weight models. The repository provides comprehensive documentation covering performance evaluations, quickstart guides for using EXAONE Deep models with Transformers, explanations of quantized EXAONE Deep weights in AWQ and GGUF formats, and instructions for running EXAONE Deep models locally using frameworks like llama.cpp and Ollama.Starting Price: Free -
4
LFM2
Liquid AI
LFM2 is a next-generation series of on-device foundation models built to deliver the fastest generative-AI experience across a wide range of endpoints. It employs a new hybrid architecture that achieves up to 2x faster decode and prefill performance than comparable models, and up to 3x improvements in training efficiency compared to the previous generation. These models strike an optimal balance of quality, latency, and memory for deployment on embedded systems, allowing real-time, on-device AI across smartphones, laptops, vehicles, wearables, and other endpoints, enabling millisecond inference, device resilience, and full data sovereignty. Available in three dense checkpoints (0.35 B, 0.7 B, and 1.2 B parameters), LFM2 demonstrates benchmark performance that outperforms similarly sized models in tasks such as knowledge recall, mathematics, multilingual instruction-following, and conversational dialogue evaluations. -
5
vLLM
vLLM
vLLM is a high-performance library designed to facilitate efficient inference and serving of Large Language Models (LLMs). Originally developed in the Sky Computing Lab at UC Berkeley, vLLM has evolved into a community-driven project with contributions from both academia and industry. It offers state-of-the-art serving throughput by efficiently managing attention key and value memory through its PagedAttention mechanism. It supports continuous batching of incoming requests and utilizes optimized CUDA kernels, including integration with FlashAttention and FlashInfer, to enhance model execution speed. Additionally, vLLM provides quantization support for GPTQ, AWQ, INT4, INT8, and FP8, as well as speculative decoding capabilities. Users benefit from seamless integration with popular Hugging Face models, support for various decoding algorithms such as parallel sampling and beam search, and compatibility with NVIDIA GPUs, AMD CPUs and GPUs, Intel CPUs, and more. -
6
Wafer
Wafer
Wafer delivers the fastest open source LLMs for enterprise through serverless and dedicated inference built for production AI workloads. Its serverless inference gives teams access to top open models with no infrastructure, no deployment overhead, and fast APIs, including GLM-5.2-Fast for low-latency inference with EAGLE speculative decoding and a per-stream throughput SLA, GLM-5.2 as a flagship model with stronger coding and reasoning capabilities, and more. Wafer’s technology uses agents that optimize inference across the stack, identifying and enhancing bottlenecks in orchestration, algorithms, serving engines, GPU kernels, and diverse hardware. It profiles the stack to see whether latency or throughput comes from scheduling, decoding, kernels, memory pressure, or hardware fit, then tries many paths and ships the measured winner. Instead of relying on a single switch or heuristic, Wafer searches model, engine, kernel, and hardware combinations.Starting Price: Free -
7
Mistral 7B
Mistral AI
Mistral 7B is a 7.3-billion-parameter language model that outperforms larger models like Llama 2 13B across various benchmarks. It employs Grouped-Query Attention (GQA) for faster inference and Sliding Window Attention (SWA) to efficiently handle longer sequences. Released under the Apache 2.0 license, Mistral 7B is accessible for deployment across diverse platforms, including local environments and major cloud services. Additionally, a fine-tuned version, Mistral 7B Instruct, demonstrates enhanced performance in instruction-following tasks, surpassing models like Llama 2 13B Chat.Starting Price: Free -
8
RunInfra
RunInfra
RunInfra turns plain English into production AI inference endpoints. Describe your use case, and the AI agent builds, optimizes, deploys, and scales it for you; no YAML, no DevOps, no GPU configuration, just chat. It is built for shipping open source AI models as production APIs, selecting compatible models, benchmarking real GPUs, applying kernel optimizations, and deploying OpenAI-compatible HTTP endpoints. RunInfra can build LLM, speech-to-text, text-to-speech, embedding, vision-language, image-generation, RAG search, document AI, transcription, AI assistant, and multi-model reasoning pipelines when the selected model and runtime support the route. Its workflow moves from description to optimization to deployment to integration; tell RunInfra what you need, let it profile real GPUs from L4 to B200, search model variants such as AWQ, GPTQ, and FP8, tune kernels with Forge, and ship an endpoint that works with OpenAI Python and JavaScript SDKs.Starting Price: $100 per month -
9
Laguna XS 2.1
Poolside
Laguna XS 2.1 is an upgraded open weight agentic coding model designed for long-horizon work on a local machine. It uses a 33-billion-parameter Mixture-of-Experts architecture with 3 billion activated parameters per token, retaining the same efficient architecture as Laguna XS.2 while improving multilingual software engineering and terminal-style task performance. The model is built to support coding agents that inspect repositories, reason through complex changes, use tools, execute commands, and continue working across extended tasks. It is served with a 256K context window, giving agents room to work with large codebases, lengthy histories, and multi-step workflows. Laguna XS 2.1 is supported by vLLM, SGLang, NVIDIA TensorRT-LLM, Hugging Face Transformers, and Ollama, with native llama.cpp support planned. It is available in BF16, FP8, INT4, and NVFP4 checkpoints, allowing developers to choose between maximum fidelity and configurations suited to tighter VRAM or compute budgets. -
10
Mirai
Mirai
Mirai is a developer-focused on-device AI infrastructure platform designed to convert, optimize, and run machine learning models directly on Apple devices with high performance and privacy. It provides a unified pipeline that enables teams to convert and quantize models, benchmark them, distribute them, and execute inference locally. It is built specifically for Apple Silicon and aims to deliver near-zero latency, zero inference cost, and full data privacy by keeping sensitive processing on the user’s device. Through its SDK and inference engine, developers can integrate AI features into applications quickly, using hardware-aware optimizations that unlock the full power of the GPU and Neural Engine. Mirai also includes dynamic routing capabilities that automatically decide whether a request should run locally or in the cloud based on latency, privacy, or workload requirements. -
11
NexaSDK
NexaSDK
Nexa SDK is a unified developer toolkit that lets you run and ship any AI model locally on virtually any device with support for NPUs, GPUs, and CPUs, offering seamless deployment without needing cloud connectivity; it provides a fast command-line interface, Python bindings, mobile (Android and iOS) SDKs, and Linux support so you can integrate AI into apps, IoT devices, automotive systems, and desktops with minimal setup and one line of code to run models, while also exposing an OpenAI-compatible REST API and function calling for easy integration with existing clients. Powered by the company’s custom NexaML inference engine built from the kernel up for optimal performance on every hardware stack, the SDK supports multiple model formats including GGUF, MLX, and Nexa’s proprietary format, delivers full multimodal support for text, image, and audio tasks (including embeddings, reranking, speech recognition, and text-to-speech), and prioritizes Day-0 support for the latest architectures. -
12
Photon
Moondream
Photon is Moondream’s official high-performance inference engine, designed to run vision-language models efficiently across cloud, desktop, and edge environments while delivering real-time performance for production AI systems. It is built as a custom inference layer tightly integrated with the Moondream model architecture, using optimized scheduling, native image processing, and purpose-built CUDA kernels to maximize speed and efficiency. This co-designed approach allows Photon to significantly reduce latency compared to traditional VLM setups, enabling responsive interactions on edge devices and real-time throughput on server-grade hardware. It supports deployment across a wide range of NVIDIA GPUs, from embedded systems like Jetson devices to high-end multi-GPU servers, making it adaptable for diverse operational needs. It includes production-ready features such as automatic batching, prefix caching, and memory-efficient attention mechanisms.Starting Price: $300 per month -
13
NVIDIA TensorRT
NVIDIA
NVIDIA TensorRT is an ecosystem of APIs for high-performance deep learning inference, encompassing an inference runtime and model optimizations that deliver low latency and high throughput for production applications. Built on the CUDA parallel programming model, TensorRT optimizes neural network models trained on all major frameworks, calibrating them for lower precision with high accuracy, and deploying them across hyperscale data centers, workstations, laptops, and edge devices. It employs techniques such as quantization, layer and tensor fusion, and kernel tuning on all types of NVIDIA GPUs, from edge devices to PCs to data centers. The ecosystem includes TensorRT-LLM, an open source library that accelerates and optimizes inference performance of recent large language models on the NVIDIA AI platform, enabling developers to experiment with new LLMs for high performance and quick customization through a simplified Python API.Starting Price: Free -
14
Qwen3.5-Plus
Alibaba
Qwen3.5-Plus is a high-performance native vision-language model designed for efficient text generation, deep reasoning, and multimodal understanding. Built on a hybrid architecture that combines linear attention with a sparse mixture-of-experts design, it delivers strong performance while optimizing inference efficiency. The model supports text, image, and video inputs and produces text outputs, making it suitable for complex multimodal workflows. With a massive 1 million token context window and up to 64K output tokens, Qwen3.5-Plus enables long-form reasoning and large-scale document analysis. It includes advanced capabilities such as structured outputs, function calling, web search, and tool integration via the Responses API. The model supports prefix continuation, caching, batch processing, and fine-tuning for flexible deployment. Designed for developers and enterprises, Qwen3.5-Plus provides scalable, high-throughput AI performance with OpenAI-compatible API access.Starting Price: $0.4 per 1M tokens -
15
Mixtral 8x7B
Mistral AI
Mixtral 8x7B is a high-quality sparse mixture of experts model (SMoE) with open weights. Licensed under Apache 2.0. Mixtral outperforms Llama 2 70B on most benchmarks with 6x faster inference. It is the strongest open-weight model with a permissive license and the best model overall regarding cost/performance trade-offs. In particular, it matches or outperforms GPT-3.5 on most standard benchmarks.Starting Price: Free -
16
LTM-2-mini
Magic AI
LTM-2-mini is a 100M token context model: LTM-2-mini. 100M tokens equals ~10 million lines of code or ~750 novels. For each decoded token, LTM-2-mini’s sequence-dimension algorithm is roughly 1000x cheaper than the attention mechanism in Llama 3.1 405B1 for a 100M token context window. The contrast in memory requirements is even larger – running Llama 3.1 405B with a 100M token context requires 638 H100s per user just to store a single 100M token KV cache.2 In contrast, LTM requires a small fraction of a single H100’s HBM per user for the same context. -
17
CodeQwen
Alibaba
CodeQwen is the code version of Qwen, the large language model series developed by the Qwen team, Alibaba Cloud. It is a transformer-based decoder-only language model pre-trained on a large amount of data of codes. Strong code generation capabilities and competitive performance across a series of benchmarks. Supporting long context understanding and generation with the context length of 64K tokens. CodeQwen supports 92 coding languages and provides excellent performance in text-to-SQL, bug fixes, etc. You can just write several lines of code with transformers to chat with CodeQwen. Essentially, we build the tokenizer and the model from pre-trained methods, and we use the generate method to perform chatting with the help of the chat template provided by the tokenizer. We apply the ChatML template for chat models following our previous practice. The model completes the code snippets according to the given prompts, without any additional formatting.Starting Price: Free -
18
Fireworks AI
Fireworks AI
Fireworks partners with the world's leading generative AI researchers to serve the best models, at the fastest speeds. Independently benchmarked to have the top speed of all inference providers. Use powerful models curated by Fireworks or our in-house trained multi-modal and function-calling models. Fireworks is the 2nd most used open-source model provider and also generates over 1M images/day. Our OpenAI-compatible API makes it easy to start building with Fireworks. Get dedicated deployments for your models to ensure uptime and speed. Fireworks is proudly compliant with HIPAA and SOC2 and offers secure VPC and VPN connectivity. Meet your needs with data privacy - own your data and your models. Serverless models are hosted by Fireworks, there's no need to configure hardware or deploy models. Fireworks.ai is a lightning-fast inference platform that helps you serve generative AI models.Starting Price: $0.20 per 1M tokens -
19
kluster.ai
kluster.ai
Kluster.ai is a developer-centric AI cloud platform designed to deploy, scale, and fine-tune large language models (LLMs) with speed and efficiency. Built for developers by developers, it offers Adaptive Inference, a flexible and scalable service that adjusts seamlessly to workload demands, ensuring high-performance processing and consistent turnaround times. Adaptive Inference provides three distinct processing options: real-time inference for ultra-low latency needs, asynchronous inference for cost-effective handling of flexible timing tasks, and batch inference for efficient processing of high-volume, bulk tasks. It supports a range of open-weight, cutting-edge multimodal models for chat, vision, code, and more, including Meta's Llama 4 Maverick and Scout, Qwen3-235B-A22B, DeepSeek-R1, and Gemma 3 . Kluster.ai's OpenAI-compatible API allows developers to integrate these models into their applications seamlessly.Starting Price: $0.15per input -
20
FriendliAI
FriendliAI
FriendliAI is a generative AI infrastructure platform that offers fast, efficient, and reliable inference solutions for production environments. It provides a suite of tools and services designed to optimize the deployment and serving of large language models (LLMs) and other generative AI workloads at scale. Key offerings include Friendli Endpoints, which allow users to build and serve custom generative AI models, saving GPU costs and accelerating AI inference. It supports seamless integration with popular open source models from the Hugging Face Hub, enabling lightning-fast, high-performance inference. FriendliAI's cutting-edge technologies, such as Iteration Batching, Friendli DNN Library, Friendli TCache, and Native Quantization, contribute to significant cost savings (50–90%), reduced GPU requirements (6× fewer GPUs), higher throughput (10.7×), and lower latency (6.2×).Starting Price: $5.9 per hour -
21
CodeT5
Salesforce
Code for CodeT5, a new code-aware pre-trained encoder-decoder model. Identifier-aware unified pre-trained encoder-decoder models for code understanding and generation. This is the official PyTorch implementation for the EMNLP 2021 paper from Salesforce Research. CodeT5-large-ntp-py is specially optimized for Python code generation tasks and employed as the foundation model for our CodeRL, yielding new SOTA results on the APPS Python competition-level program synthesis benchmark. This repo provides the code for reproducing the experiments in CodeT5. CodeT5 is a new pre-trained encoder-decoder model for programming languages, which is pre-trained on 8.35M functions in 8 programming languages (Python, Java, JavaScript, PHP, Ruby, Go, C, and C#). In total, it achieves state-of-the-art results on 14 sub-tasks in a code intelligence benchmark - CodeXGLUE. Generate code based on the natural language description. -
22
Falcon-40B
Technology Innovation Institute (TII)
Falcon-40B is a 40B parameters causal decoder-only model built by TII and trained on 1,000B tokens of RefinedWeb enhanced with curated corpora. It is made available under the Apache 2.0 license. Why use Falcon-40B? It is the best open-source model currently available. Falcon-40B outperforms LLaMA, StableLM, RedPajama, MPT, etc. See the OpenLLM Leaderboard. It features an architecture optimized for inference, with FlashAttention and multiquery. It is made available under a permissive Apache 2.0 license allowing for commercial use, without any royalties or restrictions. ⚠️ This is a raw, pretrained model, which should be further finetuned for most usecases. If you are looking for a version better suited to taking generic instructions in a chat format, we recommend taking a look at Falcon-40B-Instruct.Starting Price: Free -
23
Solar Mini
Upstage AI
Solar Mini is a pre‑trained large language model that delivers GPT‑3.5‑comparable responses with 2.5× faster inference while staying under 30 billion parameters. It achieved first place on the Hugging Face Open LLM Leaderboard in December 2023 by combining a 32‑layer Llama 2 architecture, initialized with high‑quality Mistral 7B weights, with an innovative “depth up‑scaling” (DUS) approach that deepens the model efficiently without adding complex modules. After DUS, continued pretraining restores and enhances performance, and instruction tuning in a QA format, especially for Korean, refines its ability to follow user prompts, while alignment tuning ensures its outputs meet human or advanced AI preferences. Solar Mini outperforms competitors such as Llama 2, Mistral 7B, Ko‑Alpaca, and KULLM across a variety of benchmarks, proving that compact size need not sacrifice capability.Starting Price: $0.1 per 1M tokens -
24
GigaChat 3 Ultra
Sberbank
GigaChat 3 Ultra is a 702-billion-parameter Mixture-of-Experts model built from scratch to deliver frontier-level reasoning, multilingual capability, and deep Russian-language fluency. It activates just 36 billion parameters per token, enabling massive scale with practical inference speeds. The model was trained on a 14-trillion-token corpus combining natural, multilingual, and high-quality synthetic data to strengthen reasoning, math, coding, and linguistic performance. Unlike modified foreign checkpoints, GigaChat 3 Ultra is entirely original—giving developers full control, modern alignment, and a dataset free of inherited limitations. Its architecture leverages MoE, MTP, and MLA to match open-source ecosystems and integrate easily with popular inference and fine-tuning tools. With leading results on Russian benchmarks and competitive performance on global tasks, GigaChat 3 Ultra represents one of the largest and most capable open-source LLMs in the world.Starting Price: Free -
25
LFM2.5
Liquid AI
Liquid AI’s LFM2.5 is the next generation of on-device AI foundation models designed to deliver high-performance, efficient AI inference on edge devices such as phones, laptops, vehicles, IoT systems, and embedded hardware without relying on cloud compute. It extends the previous LFM2 architecture by significantly increasing the pretraining scale and reinforcement learning stages, yielding a family of hybrid models around 1.2 billion parameters that balance instruction following, reasoning, and multimodal capabilities for real-world agentic use cases. The LFM2.5 family includes Base (for fine-tuning and customization), Instruct (general-purpose instruction-tuned), Japanese-optimized, Vision-Language, and Audio-Language variants, all optimized for fast, on-device inference under tight memory constraints and available as open-weight models deployable via frameworks like llama.cpp, MLX, vLLM, and ONNX.Starting Price: Free -
26
Pioneer
Pioneer.ai
Pioneer is an inference API built for developers who would rather ship than babysit a GPU cluster. It lets teams point an existing OpenAI, Anthropic, or other client at Pioneer, keep the same API and code, and run inference like normal while Pioneer finds where the current model falls short. It clusters production traffic by use case, surfaces where accuracy, latency, or cost can improve, then builds and routes to small specialist models automatically. Its continuous improvement loop, Adaptive Inference, mines live production failures for high-signal examples, retrains a specialist model, evaluates the new checkpoint, and promotes improvements behind the same endpoint without requiring redeployment. Pioneer supports encoder models for structured extraction tasks such as named entity recognition, text classification, structured JSON extraction, privacy filtering, and safety classification, as well as decoder models for text generation, classification, open-ended prompting, etc. -
27
LEAP
Liquid AI
The LEAP Edge AI Platform offers a full-stack on-device AI toolchain that enables developers to build edge AI applications, from model selection through inference, entirely on device. It includes a best-model search engine to find the most appropriate model for a given task and device constraint, a curated library of pre-trained model bundles ready for download, and fine-tuning tools (such as GPU-optimized scripts) for customizing models like LFM2 to specific use cases. It supports vision-enabled capabilities across iOS, Android, and laptop devices, and includes function-calling so AI models can interact with external systems via structured outputs. For deployment, LEAP provides an Edge SDK that lets developers load and query models locally, just like a cloud API, but entirely offline, and a model bundling service to package any supported model or checkpoint into a bundle optimized for edge deployment.Starting Price: Free -
28
Tensormesh
Tensormesh
Tensormesh is a caching layer built specifically for large-language-model inference workloads that enables organizations to reuse intermediate computations, drastically reduce GPU usage, and accelerate time-to-first-token and latency. It works by capturing and reusing key-value cache states that are normally thrown away after each inference, thereby cutting redundant compute and delivering “up to 10x faster inference” while substantially lowering GPU load. It supports deployments in public cloud or on-premises, with full observability and enterprise-grade control, SDKs/APIs, and dashboards for integration into existing inference pipelines, and compatibility with inference engines such as vLLM out of the box. Tensormesh emphasizes performance at scale, including sub-millisecond repeated queries, while optimizing every layer of inference from caching through computation. -
29
Overshoot
Overshoot
Overshoot provides developer infrastructure for applications that need to understand live video with low latency. Developers create a stream, publish camera, screen, or video input through LiveKit, and query frames or time segments using hosted vision-language models through an OpenAI-compatible chat-completions interface. Overshoot handles live video ingestion, model serving, routing, stream lifecycle, and multimodal preprocessing. Applications can request natural-language responses or structured JSON without operating a dedicated GPU fleet or building a custom real-time video inference stack. It is designed for robotics, physical security, gaming, sports analysis, screen understanding, industrial automation, and camera-based agents. -
30
LongLLaMA
LongLLaMA
This repository contains the research preview of LongLLaMA, a large language model capable of handling long contexts of 256k tokens or even more. LongLLaMA is built upon the foundation of OpenLLaMA and fine-tuned using the Focused Transformer (FoT) method. LongLLaMA code is built upon the foundation of Code Llama. We release a smaller 3B base variant (not instruction tuned) of the LongLLaMA model on a permissive license (Apache 2.0) and inference code supporting longer contexts on hugging face. Our model weights can serve as the drop-in replacement of LLaMA in existing implementations (for short context up to 2048 tokens). Additionally, we provide evaluation results and comparisons against the original OpenLLaMA models.Starting Price: Free -
31
Luminal
Luminal
Luminal is a machine-learning framework built for speed, simplicity, and composability, focusing on static graphs and compiler-based optimization to deliver high performance even for complex neural networks. It compiles models into minimal “primops” (only 12 primitive operations) and then applies compiler passes to replace those with device-specific optimized kernels, enabling efficient execution on GPU or other backends. It supports modules (building blocks of networks with a standard forward API) and the GraphTensor interface (typed tensors and graphs at compile time) for model definition and execution. Luminal’s core remains intentionally small and hackable, with extensibility via external compilers for datatypes, devices, training, quantization, and more. Quick-start guidance shows how to clone the repo, build a “Hello World” example, or run a larger model like LLaMA 3 using GPU features. -
32
Llama 2
Meta
The next generation of our open source large language model. This release includes model weights and starting code for pretrained and fine-tuned Llama language models — ranging from 7B to 70B parameters. Llama 2 pretrained models are trained on 2 trillion tokens, and have double the context length than Llama 1. Its fine-tuned models have been trained on over 1 million human annotations. Llama 2 outperforms other open source language models on many external benchmarks, including reasoning, coding, proficiency, and knowledge tests. Llama 2 was pretrained on publicly available online data sources. The fine-tuned model, Llama-2-chat, leverages publicly available instruction datasets and over 1 million human annotations. We have a broad range of supporters around the world who believe in our open approach to today’s AI — companies that have given early feedback and are excited to build with Llama 2.Starting Price: Free -
33
Xilinx
Xilinx
The Xilinx’s AI development platform for AI inference on Xilinx hardware platforms consists of optimized IP, tools, libraries, models, and example designs. It is designed with high efficiency and ease-of-use in mind, unleashing the full potential of AI acceleration on Xilinx FPGA and ACAP. Supports mainstream frameworks and the latest models capable of diverse deep learning tasks. Provides a comprehensive set of pre-optimized models that are ready to deploy on Xilinx devices. You can find the closest model and start re-training for your applications! Provides a powerful open source quantizer that supports pruned and unpruned model quantization, calibration, and fine tuning. The AI profiler provides layer by layer analysis to help with bottlenecks. The AI library offers open source high-level C++ and Python APIs for maximum portability from edge to cloud. Efficient and scalable IP cores can be customized to meet your needs of many different applications. -
34
Yi-Large
01.AI
Yi-Large is a proprietary large language model developed by 01.AI, offering a 32k context length with both input and output costs at $2 per million tokens. It stands out with its advanced capabilities in natural language processing, common-sense reasoning, and multilingual support, performing on par with leading models like GPT-4 and Claude3 in various benchmarks. Yi-Large is designed for tasks requiring complex inference, prediction, and language understanding, making it suitable for applications like knowledge search, data classification, and creating human-like chatbots. Its architecture is based on a decoder-only transformer with enhancements such as pre-normalization and Group Query Attention, and it has been trained on a vast, high-quality multilingual dataset. This model's versatility and cost-efficiency make it a strong contender in the AI market, particularly for enterprises aiming to deploy AI solutions globally.Starting Price: $0.19 per 1M input token -
35
Lemonfox.ai
Lemonfox.ai
Our models are deployed around the world to give you the best possible response times. Integrate our OpenAI-compatible API effortlessly into your application. Begin within minutes and seamlessly scale to serve millions of users. Benefit from our extensive scale and performance optimizations, making our API 4 times more affordable than OpenAI's GPT-3.5 API. Generate text and chat with our AI model that delivers ChatGPT-level performance at a fraction of the cost. Getting started just takes a few minutes with our OpenAI-compatible API. Harness the power of one of the most advanced AI image models to craft stunning, high-quality images, graphics, and illustrations in a few seconds.Starting Price: $5 per month -
36
ChatGLM
Zhipu AI
ChatGLM-6B is an open-source, Chinese-English bilingual dialogue language model based on the General Language Model (GLM) architecture with 6.2 billion parameters. Combined with model quantization technology, users can deploy locally on consumer-grade graphics cards (only 6GB of video memory is required at the INT4 quantization level). ChatGLM-6B uses technology similar to ChatGPT, optimized for Chinese Q&A and dialogue. After about 1T identifiers of Chinese and English bilingual training, supplemented by supervision and fine-tuning, feedback self-help, human feedback reinforcement learning and other technologies, ChatGLM-6B with 6.2 billion parameters has been able to generate answers that are quite in line with human preferences.Starting Price: Free -
37
SiliconFlow
SiliconFlow
SiliconFlow is a high-performance, developer-focused AI infrastructure platform offering a unified and scalable solution for running, fine-tuning, and deploying both language and multimodal models. It provides fast, reliable inference across open source and commercial models, thanks to blazing speed, low latency, and high throughput, with flexible options such as serverless endpoints, dedicated compute, or private cloud deployments. Platform capabilities include one-stop inference, fine-tuning pipelines, and reserved GPU access, all delivered via an OpenAI-compatible API and complete with built-in observability, monitoring, and cost-efficient smart scaling. For diffusion-based tasks, SiliconFlow offers the open source OneDiff acceleration library, while its BizyAir runtime supports scalable multimodal workloads. Designed for enterprise-grade stability, it includes features like BYOC (Bring Your Own Cloud), robust security, and real-time metrics.Starting Price: $0.04 per image -
38
Deci
Deci AI
Easily build, optimize, and deploy fast & accurate models with Deci’s deep learning development platform powered by Neural Architecture Search. Instantly achieve accuracy & runtime performance that outperform SoTA models for any use case and inference hardware. Reach production faster with automated tools. No more endless iterations and dozens of different libraries. Enable new use cases on resource-constrained devices or cut up to 80% of your cloud compute costs. Automatically find accurate & fast architectures tailored for your application, hardware and performance targets with Deci’s NAS based AutoNAC engine. Automatically compile and quantize your models using best-of-breed compilers and quickly evaluate different production settings. Automatically compile and quantize your models using best-of-breed compilers and quickly evaluate different production settings. -
39
Locai
Locai
Locai is the infrastructure and tooling to run AI off-cloud at scale. Everything runs on your own hardware, on-prem or in private cloud, with your data never leaving your estate. The platform covers the full stack. Locai Link is the inference runtime: deploy LLMs, audio transcription, video processing and image detection on any hardware, from workstations to servers to edge devices. Locai Control is the orchestration plane: manage models, updates and policy across whole fleets of machines from one place, with output from one model flowing into the next for true multimodal data fusion. Locai Workspace gives teams a private AI workspace on top: chat, documents and assistants powered entirely by models running inside your organisation. Locai is hardware agnostic and model flexible, running any GGUF, ONNX or TFLite model. Built in the UK for regulated organisations in financial services, nuclear, defence and critical infrastructure, and free to start. -
40
Mu
Microsoft
Mu is a 330-million-parameter encoder–decoder language model designed to power the agent in Windows settings by mapping natural-language queries to Settings function calls, running fully on-device via NPUs at over 100 tokens per second while maintaining high accuracy. Drawing on Phi Silica optimizations, Mu’s encoder–decoder architecture reuses a fixed-length latent representation to cut computation and memory overhead, yielding 47 percent lower first-token latency and 4.7× higher decoding speed on Qualcomm Hexagon NPUs compared to similar decoder-only models. Hardware-aware tuning, including a 2/3–1/3 encoder–decoder parameter split, weight sharing between input and output embeddings, Dual LayerNorm, rotary positional embeddings, and grouped-query attention, enables fast inference at over 200 tokens per second on devices like Surface Laptop 7 and sub-500 ms response times for settings queries. -
41
Sanctum
Sanctum
Sanctum is a private, local AI assistant that lets users run and interact with full-featured open source LLMs locally on their own device. It is built as a private Sanctum for AI, where data is encrypted, secure, and never leaves the user’s computer. It makes it easy to run AI locally, with an easy-to-download desktop solution that lets users run large language models on a Mac instantly without complicated installs, and after downloading, no internet connection is required. Sanctum is designed around privacy-first local processing: with on-device encryption and processing, users maintain complete privacy and control. Its Hugging Face integration gives access to thousands of GGUF models directly through Sanctum, making it possible to check compatibility, download models, and start using them on a PC or Mac. Sanctum also supports private PDF workflows, allowing users to chat with PDFs, ask questions, and summarize files in a secure environment. -
42
GLM-OCR
Z.ai
GLM-OCR is a multimodal optical character recognition model and open source repository that provides accurate, efficient, and comprehensive document understanding by combining text and visual modalities into a unified encoder–decoder architecture derived from the GLM-V family. Built with a visual encoder pre-trained on large-scale image–text data and a lightweight cross-modal connector feeding into a GLM-0.5B language decoder, the model supports layout detection, parallel region recognition, and structured output for text, tables, formulas, and complicated real-world document formats. It introduces Multi-Token Prediction (MTP) loss and stable full-task reinforcement learning to improve training efficiency, recognition accuracy, and generalization, achieving state-of-the-art benchmarks on major document understanding tasks.Starting Price: Free -
43
LLaMA-Factory
hoshi-hiyouga
LLaMA-Factory is an open source platform designed to streamline and enhance the fine-tuning process of over 100 Large Language Models (LLMs) and Vision-Language Models (VLMs). It supports various fine-tuning techniques, including Low-Rank Adaptation (LoRA), Quantized LoRA (QLoRA), and Prefix-Tuning, allowing users to customize models efficiently. It has demonstrated significant performance improvements; for instance, its LoRA tuning offers up to 3.7 times faster training speeds with better Rouge scores on advertising text generation tasks compared to traditional methods. LLaMA-Factory's architecture is designed for flexibility, supporting a wide range of model architectures and configurations. Users can easily integrate their datasets and utilize the platform's tools to achieve optimized fine-tuning results. Detailed documentation and diverse examples are provided to assist users in navigating the fine-tuning process effectively.Starting Price: Free -
44
Oxlo.ai
Oxlo.ai
Oxlo.ai is a privacy-first inference stack for agents, built to run frontier-class open-source models with unlimited agentic tool calls, secure failover, and zero data retention or training. It gives developers request-based access to curated open models through a unified HTTP API designed for predictable usage, low-latency inference, and clean integration into production systems. Teams can call models through OpenAI-compatible endpoints, switch from another provider by changing the base URL and API key, and keep support for streaming, function calling, JSON mode, vision models, embeddings, and image generation. Oxlo.ai supports more than 40 models across text, chat, reasoning, coding, image generation, audio, embeddings, computer vision, vision-language, speech-to-text, text-to-speech, long-context, and detection workflows.Starting Price: $80 per month -
45
DeepSeek-V3.2-Exp
DeepSeek
Introducing DeepSeek-V3.2-Exp, our latest experimental model built on V3.1-Terminus, debuting DeepSeek Sparse Attention (DSA) for faster and more efficient inference and training on long contexts. DSA enables fine-grained sparse attention with minimal loss in output quality, boosting performance for long-context tasks while reducing compute costs. Benchmarks indicate that V3.2-Exp performs on par with V3.1-Terminus despite these efficiency gains. The model is now live across app, web, and API. Alongside this, the DeepSeek API prices have been cut by over 50% immediately to make access more affordable. For a transitional period, users can still access V3.1-Terminus via a temporary API endpoint until October 15, 2025. DeepSeek welcomes feedback on DSA via its feedback portal. In conjunction with the release, DeepSeek-V3.2-Exp has been open-sourced: the model weights and supporting technology (including key GPU kernels in TileLang and CUDA) are available on Hugging Face.Starting Price: Free -
46
Oumi
Oumi
Oumi is a fully open source platform that streamlines the entire lifecycle of foundation models, from data preparation and training to evaluation and deployment. It supports training and fine-tuning models ranging from 10 million to 405 billion parameters using state-of-the-art techniques such as SFT, LoRA, QLoRA, and DPO. The platform accommodates both text and multimodal models, including architectures like Llama, DeepSeek, Qwen, and Phi. Oumi offers tools for data synthesis and curation, enabling users to generate and manage training datasets effectively. For deployment, it integrates with popular inference engines like vLLM and SGLang, ensuring efficient model serving. The platform also provides comprehensive evaluation capabilities across standard benchmarks to assess model performance. Designed for flexibility, Oumi can run on various environments, from local laptops to cloud infrastructures such as AWS, Azure, GCP, and Lambda.Starting Price: Free -
47
DBRX
Databricks
Today, we are excited to introduce DBRX, an open, general-purpose LLM created by Databricks. Across a range of standard benchmarks, DBRX sets a new state-of-the-art for established open LLMs. Moreover, it provides the open community and enterprises building their own LLMs with capabilities that were previously limited to closed model APIs; according to our measurements, it surpasses GPT-3.5, and it is competitive with Gemini 1.0 Pro. It is an especially capable code model, surpassing specialized models like CodeLLaMA-70B in programming, in addition to its strength as a general-purpose LLM. This state-of-the-art quality comes with marked improvements in training and inference performance. DBRX advances the state-of-the-art in efficiency among open models thanks to its fine-grained mixture-of-experts (MoE) architecture. Inference is up to 2x faster than LLaMA2-70B, and DBRX is about 40% of the size of Grok-1 in terms of both total and active parameter counts. -
48
Geekflare Chat
Geekflare
Geekflare Chat is an all-in-one AI platform that bundles the world’s most powerful models from OpenAI, Anthropic Claude, and Google Gemini into a collaborative workspace. By consolidating OpenAI, Anthropic, and Google into one interface, Geekflare Chat removes the friction of modern AI. Teams can use the Multi-Model Comparison tool to evaluate responses from GPT-5.4, Claude 4.5, and Gemini 3.1 Pro side-by-side. Collaboration is built natively into the platform, allowing teams to share workspaces, build a centralized AI Knowledge Base, and standardize outputs with a shared Prompt Library. Start chatting for free, or upgrade to our Business Plan to give your entire team the AI advantage they need to move faster for just $29/month.Starting Price: $9/month -
49
Mercury 2
Inception
Mercury 2 is the first reasoning model fast enough to pick up the phone, a reasoning diffusion language model built for real-time voice agents. Instead of making callers wait through seconds of dead air while an autoregressive model generates thinking tokens one by one, Mercury 2 uses a diffusion large language model architecture to generate tokens in parallel, decoding 1000+ tokens per second on standard NVIDIA GPUs. That speed is fast enough to run a full reasoning pass and start speaking within the latency budget of a natural conversation, reducing the cost of reasoning from seconds of silence to roughly 300 milliseconds. Mercury models work by corrupting clean text into noise, then training a standard Transformer to reverse the process and predict clean text across all positions simultaneously. Because each denoising pass touches many tokens, generation uses the GPU more efficiently than one-token-at-a-time decoding, making custom-silicon-like speed possible on NVIDIA H100s. -
50
Phi-4-mini-flash-reasoning
Microsoft
Phi-4-mini-flash-reasoning is a 3.8 billion‑parameter open model in Microsoft’s Phi family, purpose‑built for edge, mobile, and other resource‑constrained environments where compute, memory, and latency are tightly limited. It introduces the SambaY decoder‑hybrid‑decoder architecture with Gated Memory Units (GMUs) interleaved alongside Mamba state‑space and sliding‑window attention layers, delivering up to 10× higher throughput and a 2–3× reduction in latency compared to its predecessor without sacrificing advanced math and logic reasoning performance. Supporting a 64 K‑token context length and fine‑tuned on high‑quality synthetic data, it excels at long‑context retrieval, reasoning tasks, and real‑time inference, all deployable on a single GPU. Phi-4-mini-flash-reasoning is available today via Azure AI Foundry, NVIDIA API Catalog, and Hugging Face, enabling developers to build fast, scalable, logic‑intensive applications.