Open-Weight & Local Models

MiMo-V2.6-Pro

Xiaomi's flagship foundation model, built at a scale of over 1T parameters for the most demanding workloads. Natively multimodal across text, image, video, and audio input, with reasoning, tool use, and a 1M-token context window; weights are published on Hugging Face.

MiMo-V2.6-Flash

Xiaomi's open-source mixture-of-experts foundation model with 309B total parameters and 15B activated per token, using a hybrid attention mechanism for efficient long-context inference. Multimodal across text, image, video, and audio input, with reasoning, tool use, and a 1M-token context window.

Ternary Bonsai 2 27B

PrismML's open-weight 27B-parameter reasoning model derived from Qwen 3.8 27B and shrunk with ternary compression. Supports coding, mathematics, tool calling, and image understanding with a 262K-token context window.

Schematron V2 Small

Inference.net's open-weight 3B-parameter HTML-to-JSON extraction model, prioritizing extraction quality for complex schemas and long pages. Extraction instructions are supplied through a JSON schema rather than a prompt. 128K-token context window.

Schematron V2 Turbo

Inference.net's open-weight 3B-parameter HTML-to-JSON extraction model, prioritizing throughput for high-volume extraction workloads. Extraction instructions are supplied through a JSON schema rather than a prompt. 128K-token context window.

Ling 3.0 Flash VL

InclusionAI's open-weight vision-language model built on Ling 3.0 Flash (124B total, 5.5B active MoE), strengthening language capabilities while adding native visual perception and visual reasoning over images and video. 262K-token context window.

DeepSeek V4.1 Flash

DeepSeek's open-weight sparse mixture-of-experts model and the first built on the company's Causal Encoder-Decoder (CED) architecture, activating 8B parameters on input and 16B on output. Supports image input, reasoning, and tool use with a 1M-token context window at very low per-token cost.

Nex-N2.5-Mini

The smaller member of Nex AGI's open-weight Nex-N2.5 agentic family, aimed at agentic coding within a visual feedback loop at lower cost. Supports image input, reasoning, and a 262K-token context window.

Nex-N2.5-Pro

Nex AGI's open-weight agentic model built to turn goals into working, verified outcomes. Its core strength is agentic coding within a visual feedback loop, exploring codebases and implementing multi-file changes, with image input, tool use, and a 262K-token context window.

Hy4 Preview

Tencent Hunyuan's fourth-generation open-weight flagship preview model, a Mixture-of-Experts system with 770B total parameters and 49B active parameters per token. It targets real-world productivity tasks across coding, office work, scientific research, and multi-step tool workflows, supports function calling, reasoning modes, and a context window of about one million tokens.

Qwen3.8-Flash-Next

Alibaba's open-weight multimodal preview of the architecture being developed for Qwen4. The sparse mixture-of-experts model has a 125B-parameter backbone with 6B parameters active per token plus 51B parameters of N-gram embeddings. It combines Gated DeltaNet with Qwen Sparse Attention, supports 262,144 tokens natively and up to one million tokens with YaRN, and targets efficient coding, office, visual, and long-context workloads.

GLM-5.3-Flash

Z.ai's first natively multimodal model in the GLM-5 family, designed for efficient coding, agentic work, visual reasoning, and long-context tasks. It uses a 320B-parameter sparse mixture-of-experts architecture with 18B active parameters, hybrid linear and sparse attention, and a context window of up to one million tokens. The model weights are released under the MIT license for local deployment with frameworks including SGLang and vLLM.

Granite Speech 5.0 470M TurboCTC

IBM's compact 470M parameter English automatic speech recognition model from the Granite Speech 5.0 family. Uses a conformer acoustic encoder with CTC, non-autoregressive greedy decoding, and high-throughput transcription for edge and enterprise speech workloads. Released under Apache 2.0.

Granite 4.2 8B

IBM's balanced dense decoder-only reasoning language model from the Granite 4.2 family. Designed for enterprise agentic workflows, coding, tool calling, instruction following, multilingual dialog across 12 tested languages, and native thinking modes with a 128K-token context window. Released under Apache 2.0.

Granite 4.2 30B

IBM's flagship dense decoder-only reasoning language model from the Granite 4.2 family. Built for complex reasoning, coding, agentic tool use, instruction following, and multilingual dialog across 12 tested languages, with native 128K context and long-context extension up to 512K tokens. Released under Apache 2.0.

Granite 4.2 3B

IBM's compact dense decoder-only reasoning language model from the Granite 4.2 family. Supports native thinking modes, reasoning-augmented tool calling, code generation, multilingual dialog across 12 tested languages, and a 128K-token context window. Released under Apache 2.0.

Granite Speech 5.0 470M TurboCTC NC

IBM's compact 470M parameter non-commercial English automatic speech recognition model from the Granite Speech 5.0 family. Uses a conformer acoustic encoder with CTC, non-autoregressive greedy decoding, and high-throughput transcription for research and non-commercial speech workloads.

Hy-MT2-30B-A3B

Tencent's flagship open-weight translation model in the Hy-MT2 family. Supports 33 language pairs and five Chinese dialect and minority-language pairs, with workflows for structured, contextual, glossary-based, and style-guided translation.

Hy-MT2-1.8B

Tencent's compact open-weight 1.8B-parameter translation model. Supports 33 language pairs and five Chinese dialect and minority-language pairs, with workflows for structured, contextual, glossary-based, and style-guided translation.

Hy-MT2-7B

Tencent's open-weight 7B-parameter translation model. Supports 33 language pairs and five Chinese dialect and minority-language pairs, with workflows for structured, delimiter-based, contextual, glossary-based, and style-guided translation.

Qwen 3.8 27B

Alibaba's open-weight 27B dense vision-language model in the Qwen 3.8 family. It builds on the Qwen 3.6 27B line with stronger coding and office productivity capabilities across text and visual modalities, supports one million tokens of context through hosted APIs, and is suited to multimodal assistants, document work, coding, and long-running agent tasks.

Dots3-Note Preview

Dots Studio's open-weight mixture-of-experts model with 16B active parameters out of 280B total, the lightest model in the Dots 3 family. Supports image input, reasoning, tool use, and a 512K-token context window.

GLM-5.3

Z.ai's (formerly Zhipu AI) coding and agentic upgrade built on the same base model as GLM-5.2, with all gains coming from extended post-training (more executable environments, longer-horizon tasks, more RL compute). Claims the strongest open-weight coding performance to date, with large jumps on Terminal-Bench 3.0, DeepSWE v1.1, SWE-Marathon v1.1, and cybersecurity benchmarks like CyberGym and ExploitBench. Supports low, high, and max reasoning effort levels with a 1M-token context route.

Qwen 3.8 2.4T-A95B

Alibaba's open-weight Max-class Qwen 3.8 model, a sparse Mixture-of-Experts system with 2.4T total parameters and about 95B active parameters per token. The text model targets coding, research, complex reasoning, and long-horizon agentic workflows with a native 262K context window that can be extended to about one million tokens and is served by hosted providers with 1M context.

Nemotron 3.5 Lightning

NVIDIA's open mixture-of-experts model with 3B active parameters out of 30B total, suited for high-throughput agentic workloads and specialized tasks. Supports reasoning and tool use with a 262K-token context window.

LFM2.5-2.6B

Liquid AI's compact open-weight reasoning model, suited for agent workflows, data extraction, RAG, and long-context processing. Liquid advises against using it for agentic coding. 65K-token context window.

Muse Glimmer 30B

A dense, open-weight 30B multimodal model from Meta Superintelligence Labs, distilled from Muse Spark and optimized for autonomous agents on consumer hardware. Supports image input, reasoning, and tool use with a 131K-token context window.

Inkling Small

An open-weight multimodal mixture-of-experts model from Thinking Machines Lab with 12B active parameters out of 276B total, positioned as the smaller, more efficient member of the Inkling family. Accepts text, image, and audio input with reasoning, tool use, and a 524K-token context window.

Ling 3.0 Flash

InclusionAI's open-weight 124B-parameter mixture-of-experts model with about 5.1B parameters activated per token, designed for token efficiency and production-scale agentic inference. Supports reasoning, tool use, and a 262K-token context window.

Laguna S 2.1

Poolside's open-weight coding agent model with 118B total and 8B active parameters, scoring 70.2% on Terminal-Bench 2.1. Built for agentic software engineering with reasoning, tool use, and a 1M-token context window.

Motif 3 Beta

A large-scale sparse Mixture-of-Experts language model built from the ground up by South Korea's Motif Technologies with a fully in-house, proprietary design (Grouped Differential Latent Attention, per-expert Grouped PolyNorm, modified mHC). Has roughly 314B total parameters with about 13B active per token (384 routed experts, top-8, plus one shared expert) and a native 256K context window (262,144 tokens). Multilingual and general-purpose, with reasoning and tool-calling support in vLLM serving. This is an intermediate preview/beta checkpoint, released with open weights (bfloat16) for non-commercial research use ahead of the final release.

LongCat 2.0

Meituan's open-weight sparse mixture-of-experts language model with 48B active parameters out of 1.6T total. Suited for coding, repository-level changes, long-horizon problem solving, and agentic work, with reasoning, tool use, and a 1M-token context window.

Inkling

Thinking Machines Lab's open-weights general-purpose multimodal Mixture-of-Experts model with 975B total parameters and 41B active parameters. Inkling accepts text, image, and audio inputs, produces text, and is designed for agentic and tool-use systems, coding assistants, chatbots, and retrieval-augmented generation.

Kimi K3

Moonshot AI's flagship Kimi model for frontier intelligence, agentic coding, knowledge work, and deep reasoning. Kimi K3 supports a 1-million-token context window for long-running software engineering and research workflows.

Hy3

Tencent's open-weight 295B-parameter mixture-of-experts model (21B active, 192 experts with top-8 routing) built for reasoning, agentic workflows, and real-world production use, with configurable reasoning effort, tool use, and a 262K-token context window.

Laguna XS 2.1

Poolside's open-weight coding agent model in the 33B-A3B class and a step forward from Laguna XS.2. Built for efficient agentic software engineering with reasoning, tool use, and a 262K-token context window.

Sarvam-1

Sarvam AI's compact 2B-parameter language model built from the ground up for Indian languages. Provides best-in-class performance across 10 Indic languages (bn, gu, hi, kn, ml, mr, or, pa, ta, te) alongside English, outperforming larger general-purpose models like Gemma-2-2B and Llama-3.2-3B thanks to careful data curation and an efficient Indic tokenizer. Edge-deployable.

Sarvam-105B

Sarvam AI's sovereign 105B-parameter Mixture-of-Experts model activating ~9B parameters per token, with a 128K-token context window. Trained on 12 trillion tokens across 22 Indian languages using 128 sparse experts with Multi-head Latent Attention and a custom low-fertility Indic tokenizer. Wins the majority of pairwise comparisons on Indian-language and STEM benchmarks.

Sarvam-M

Sarvam AI's 24B-parameter instruction-tuned model derived from Mistral-Small-3.1-24B, post-trained on English plus eleven major Indic languages (bn, hi, kn, gu, mr, ml, or, pa, ta, te). Delivers large relative gains on Indian-language, math, and programming benchmarks over its base model, with a hybrid reasoning mode for complex tasks.

GLM-5.2

Z.ai's (formerly Zhipu AI) flagship open-weight coding model with a 1M-token context window. Mixture-of-Experts architecture with 753B total parameters and ~40B active per request, featuring two cost-balancing reasoning modes. Tops several coding benchmarks while remaining a fraction of the cost of comparable proprietary frontier models. MIT-licensed weights.

Command A+

Cohere's enterprise flagship model building on Command A with stronger reasoning, agentic tool use, and multilingual performance across 23 languages. Optimized for secure, high-throughput RAG, retrieval, and long-horizon agent workflows in regulated environments, with private and on-premise deployment options.

Sarvam-30B

Sarvam AI's 30B-parameter Mixture-of-Experts reasoning model trained from scratch with only 2.4B active parameters per token. Optimized for real-time deployment and Indian languages, delivering strong reasoning, coding, and conversational performance while remaining efficient to serve. Open-weights.

Kimi K2.7 Code

Moonshot AI's latest open-source, coding-focused model in the Kimi K2 family, built to complete end-to-end programming tasks reliably over long contexts. A 1-trillion-parameter model that cuts reasoning token usage by roughly 30% versus K2.6 while improving coding and agent performance — +21.8% on Kimi Code Bench v2, +11.0% on Program Bench, and +31.5% on MLS Bench Lite for multi-language support. Released under a Modified MIT License and available via Kimi APIs and Hugging Face.

DiffusionGemma

Google DeepMind's experimental diffusion-based member of the Gemma 4 open model family. Unlike autoregressive models that generate text one token at a time, DiffusionGemma denoises a canvas of placeholder tokens to produce up to 256 tokens in parallel, finalizing output in one block. A Mixture-of-Experts model with 26B total parameters and 3.8B active per inference, delivering roughly 4x the throughput of similarly sized autoregressive Gemma models on local hardware. Excels at non-linear tasks like in-line editing, molecular sequencing, mathematical graphing, and self-correcting puzzles.

Gemma 4 12B

Google's medium-size open-weight model with 12 billion parameters from the Gemma 4 family. Encoder-free unified multimodal architecture that natively processes text, image, audio, and video inputs without dedicated encoders. Features a 256K context window and supports 140+ languages. First medium-sized model capable of natively ingesting audio. Suitable for local deployment on GPUs with 16GB VRAM.

Nemotron 3 Ultra

NVIDIA's flagship open 550B-parameter Mixture-of-Experts model with 55B active parameters, built for frontier reasoning and orchestration in long-running agentic systems. Features hybrid Mamba-Transformer architecture, LatentMoE routing, multi-token prediction, and NVFP4 precision for 5x higher throughput. Achieves 30% lower cost-to-task-completion on agentic benchmarks. Supports 1M+ token context window with 95% accuracy on Ruler@1M.

Snowflake Arctic

Snowflake's enterprise-focused open LLM with 480B total parameters using a fine-grained MoE architecture with only 17B active parameters per input. Apache 2.0 licensed, excels at SQL generation, coding, and enterprise intelligence tasks with breakthrough training efficiency.

Falcon 3 10B

TII's open-source 10B parameter model from the Falcon 3 family. Achieved number one position on Hugging Face's LLM leaderboard in its size category, outperforming Meta's Llama variants and other models under 13B parameters.

Jamba Large 1.7

AI21's latest hybrid SSM-Transformer model with Mixture-of-Experts architecture. Features a 256K context window, improved grounding and instruction-following. 94B total parameters with 398B active, optimized for enterprise long-context tasks.

Falcon-H1

TII's hybrid Mamba-Transformer model that outperforms comparable offerings from Meta's Llama and Alibaba's Qwen in the 30-70B parameter range. Designed for real-world AI on everyday devices and resource-limited settings with state-of-the-art efficiency.

Ring-2.6-1T

InclusionAI's (Ant Group) trillion-parameter open-weights reasoning model with 63B active parameters per token. Built for real-world agent workflows with adaptive reasoning-effort modes. Features hybrid linear and MLA attention architecture with MIT license.

StableLM 2 12B

Stability AI's 12.1 billion parameter decoder-only language model pre-trained on 2 trillion tokens of diverse multilingual and code datasets. Supports multiple languages and offers strong performance for its compact size with instruction-tuned chat variant available.

DBRX

Databricks' open-source 132B parameter Mixture-of-Experts transformer model with 36B active parameters per input. Released under Databricks Open Model License, optimized for enterprise workloads including SQL generation and coding tasks.

Alloma 8B Instruct

Uzbek LLM Lab's 8B parameter instruction-tuned model optimized for the Uzbek language. Built on Llama architecture with a custom tokenizer averaging 1.7 tokens per Uzbek word versus 3.5 in original Llama, enabling 2x faster inference. Trained on 3.6B tokens with 4096 context length.

K2 Think

A 32 billion parameter open-weights reasoning model by LLM360/MBZUAI, built on Qwen2.5-32B. Trained with reinforcement learning and verifiable rewards for long chain-of-thought reasoning, agentic planning, and complex problem solving in math, science, and code.

MiniCPM-V 4.6

Ultra-efficient multimodal language model from OpenBMB built on SigLIP2-400M and Qwen3.5-0.8B (~1B parameters). Supports single-image, multi-image, and video understanding with mixed 4x/16x visual token compression. Designed for edge deployment on iOS, Android, and HarmonyOS.

Granite 4.1 8B

IBM's dense decoder-only 8B parameter language model from the Granite 4.1 family. Supports 131K-token context, tool calling, RAG, code generation with fill-in-the-middle, text summarization, classification, and extraction across 12 languages. Released under Apache 2.0.

Granite 4.1 30B

IBM's largest dense decoder-only 30B parameter language model from the Granite 4.1 family. Trained on approximately 15T tokens with long-context extension up to 512K tokens. Supports tool calling, RAG, code generation, multilingual tasks across 12 languages. Released under Apache 2.0.

DeepSeek V4 Flash

DeepSeek's efficient V4 model with 284B total parameters (13B activated). Optimized for speed and cost-efficiency while maintaining strong performance. Supports 1M token context window.

DeepSeek V4 Pro

DeepSeek's flagship V4 model with 1.6T total parameters (49B activated). MoE architecture supporting 1M token context. Closes the gap with frontier proprietary models on reasoning and coding benchmarks.

MiMo-V2.5-Pro

Xiaomi's flagship 1.02T-parameter Mixture-of-Experts model with 42B active parameters, built on a hybrid-attention architecture with 3-layer Multi-Token Prediction. Designed for complex agentic tasks, software engineering, and long-horizon instruction following with a 1M-token context window.

Qwen 3.6 27B

Alibaba's dense 27B parameter model that outperforms its own 397B MoE predecessor on agentic coding benchmarks. Strong multilingual and reasoning capabilities released under Apache 2.0.

Hy3 Preview

Tencent's flagship open-weight Mixture-of-Experts model from the Hunyuan family with 295B total parameters and 21B active. Integrates fast and slow thinking modes with configurable reasoning effort. Designed for agentic workflows, cross-file code refactoring, long-document analysis, and multi-step tool use.

Qwen 3.6 35B-A3B

Alibaba's efficient Mixture-of-Experts model with 35B total parameters and 3B active per token. Frontier-level agentic coding performance with 73.4% on SWE-bench Verified and 92.7 on AIME 2026. Released under Apache 2.0.

Gemma 4 26B

Google's high-performance open-weight dense model with 26 billion parameters from the Gemma 4 family. Supports multimodal inputs including text and images with a 256K extended context window. Strong reasoning and code generation capabilities with all parameters active per forward pass.

Gemma 4 E4B

Google's efficient 4 billion parameter variant from the Gemma 4 family. Designed for resource-constrained environments while maintaining strong text generation quality. Text-only model with a 32K context window, balancing performance and efficiency.

Gemma 4 31B

Google's flagship open-weight dense model with 31 billion parameters from the Gemma 4 family. All parameters active per forward pass with top-tier performance on reasoning benchmarks including AIME 2026 and MMLU Pro. Supports vision and extended 256K context window.

Gemma 4 E2B

Google's efficient 2 billion parameter variant from the Gemma 4 family. Optimized for on-device and edge deployments with minimal resource requirements. Text-only model with a 32K context window, suitable for lightweight chat and completion tasks.

Gemma 4 31B

Google's flagship open-weight dense model with 31B parameters. All parameters active per forward pass. Ranks among top open models with strong performance on AIME 2026 (89.2%) and MMLU Pro (85.2%). Supports vision and extended context.

GPT-OSS 120B

OpenAI's first open-weight large model with 120 billion parameters. Released under Apache 2.0 license, offering strong performance on reasoning and coding tasks while being fully self-hostable.

GPT-OSS 20B

OpenAI's compact open-weight model with 20 billion parameters. Released under Apache 2.0 license, designed for efficient deployment on consumer hardware while maintaining strong coding and reasoning capabilities.

Nemotron 3 Super 120B

NVIDIA's open hybrid Mamba-Transformer MoE model with 120B total parameters (12B active). Features 1M token context window and excels at agentic reasoning, coding, planning, and tool calling.

Qwen 3.6

Alibaba's latest Qwen model with enhanced reasoning, multilingual capabilities, and improved instruction following. Features strong performance on coding, math, and general knowledge benchmarks.

Mistral Small 4

Mistral AI's efficient hybrid model unifying instruct, reasoning, and coding in a single model. Open-weight under Apache 2.0 with strong performance for its size class.

Tiny Aya

Compact multilingual language model from Cohere For AI with 3.35B parameters, optimized for efficient and balanced multilingual representation across 70+ languages including many lower-resourced ones. Designed for edge deployment without cloud dependency. Trained on 64 NVIDIA H100 GPUs with specialized regional variants available (Global, Earth, Fire).

DeepSeek V4

DeepSeek's fourth-generation model with improved mixture-of-experts architecture, enhanced reasoning and coding capabilities, and stronger multilingual performance. Competitive with frontier proprietary models.

GigaChat 3.1 Lightning

Sber's compact Mixture-of-Experts model with 10B total parameters and 1.8B active. Designed for fast multilingual assistant workloads, reasoning, code, function calling, and product-style deployment on edge devices.

Mistral Large 3

Mistral AI's largest open-weight model with 41B active parameters (675B total MoE). State-of-the-art general-purpose multimodal model with 256K context window and powerful agentic capabilities. Released under Apache 2.0.

GigaChat 3.1 Ultra

Sber's flagship large-scale Mixture-of-Experts model with 702B total parameters and 36B active. Designed for multilingual assistant workloads, reasoning, code generation, tool use, and large-cluster deployment. Open-weight release.

Devstral 2

Mistral AI's frontier code agents model designed for solving software engineering tasks. Open-weight model optimized for agentic coding workflows and complex development tasks.

GLM-4.7

Zhipu AI's multilingual agentic coding model with strong reasoning, tool use, and UI generation capabilities. Predecessor to GLM-5.1 with competitive performance on coding benchmarks.

AlemLLM

Kazakhstan's flagship Mixture-of-Experts language model developed by Astana Hub with technical support from 01.AI. Features 247B total parameters with 22B active per token, achieving state-of-the-art results on Kazakh, Russian, and English benchmarks. Outperforms GPT-4o on Kazakh language tasks.

Trendyol LLM 8B T1

Turkish-optimized 8B chat model developed by Trendyol, Turkey's largest e-commerce platform. Built on Qwen3-8B and fine-tuned on large-scale Turkish e-commerce datasets. Features advanced chain-of-thought reasoning in Turkish with dual operation modes (/think and /no_think), strong instruction following, summarization, coding, and attribute extraction for catalogue enrichment. English reasoning capabilities are preserved alongside Turkish.

YandexGPT 5 Lite

Yandex's compact 8B parameter language model trained on 15T tokens of primarily Russian and English text. Features 32K context window with strong performance on web, code, and mathematics tasks. Open-weight release.

Nemotron Nano 9B v2

NVIDIA's compact 9B parameter model trained from scratch for both reasoning and non-reasoning tasks. Generates reasoning traces before final responses. Efficient for edge and on-device deployment.

WiroAI Turkish LLM 9B

Turkish-specialized 9B language model developed by WiroAI, built on Google's Gemma 2 architecture. Fine-tuned with Supervised Fine-Tuning (SFT) on over 500,000 carefully curated high-quality Turkish instructions, specifically adapted to Turkish culture and local context. Demonstrates superior performance on Turkish language processing tasks including conversation, reasoning, and instruction following.

Kumru 7B

Turkish large language model developed by VNGRS, pre-trained from scratch on 500 GB of Turkish corpora (300B tokens). Decoder-only architecture with a custom tokenizer optimized for Turkish, supporting code, math, and chat. Outperforms significantly larger multilingual models on the Cetvel Turkish benchmark. Available for on-premise enterprise deployment.

Llama 4 Scout

Meta's efficient MoE model with 17B active parameters (109B total, 16 experts). Supports up to 10M token context — the longest of any production model. Strong performance on reasoning and multilingual tasks.

Llama 4 Maverick

Meta's quality-focused MoE model with 17B active parameters (400B total, 128 experts). Targets quality-critical tasks with benchmark scores competitive with GPT-4o and Gemini 2.5 Pro.

Qwen3 32B

Alibaba's Qwen3 32B dense language model with strong reasoning and multilingual capabilities, supporting function calling and code generation across diverse tasks.

Qwen3 235B

Alibaba's Qwen3 235B mixture-of-experts model delivering frontier-level performance with advanced reasoning, function calling, and code generation capabilities at massive scale.

Qwen3 Coder

Alibaba's Qwen3 Coder model optimized for software development tasks including code generation, debugging, code review, and technical documentation with strong multilingual programming support.

Gemma 3 27B

Google's largest open-weight model in the Gemma 3 family with 27 billion parameters. Supports multimodal inputs including text and images with a 128K context window. Delivers strong performance across reasoning, code generation, and vision tasks, competitive with larger proprietary models.

Gemma 3 4B

Google's compact open-weight model with 4 billion parameters from the Gemma 3 family. Supports multimodal inputs including text and images with a 128K context window. Balances efficiency and capability for vision and language tasks.

Gemma 3 12B

Google's mid-size open-weight model with 12 billion parameters from the Gemma 3 family. Supports multimodal inputs including text and images with a 128K context window. Strong performance on reasoning and code generation tasks at moderate compute cost.

Gemma 3 1B

Google's lightweight open-weight model with 1 billion parameters from the Gemma 3 family. Designed for on-device and resource-constrained deployments. Supports text-only tasks with a 32K context window. Efficient for chat and basic completion workloads.

Mistral Small 3.1

Mistral AI's Small 3.1 model with 24B parameters offering efficient multimodal capabilities including vision, function calling, and code generation with a large 128K context window.

QwQ 32B

Alibaba's QwQ 32B reasoning-focused model designed for complex problem solving, mathematical reasoning, and step-by-step logical analysis with strong chain-of-thought capabilities.

Command A

Cohere's flagship 111B parameter model optimized for demanding enterprises requiring fast, secure, and high-quality AI. Excels at RAG, tool use, and multilingual tasks with strong reasoning capabilities.

Phi-4 Mini

Microsoft's Phi-4 Mini model with 3.8B parameters providing lightweight yet capable language understanding and code generation, optimized for resource-constrained deployments with a large 128K context window.

DeepSeek R1

DeepSeek's reasoning-focused model trained with reinforcement learning for complex multi-step reasoning. Excels at math, science, and coding problems requiring chain-of-thought reasoning.

Codestral

Mistral AI's cutting-edge code generation model specializing in low-latency, high-frequency tasks such as fill-in-the-middle (FIM), code completion, correction, and test generation. Features efficient architecture with 2x faster generation than its predecessor.

ISSAI KazLLM 1.0 70B

Large language model developed by ISSAI (Nazarbayev University) customized from Llama 3.1 70B to improve helpfulness of responses in the Kazakh language. Part of Kazakhstan's initiative to ensure the country benefits from generative AI advancements.

Llama 3.3 70B Instruct

Meta's flagship open-weight model with 70 billion parameters. Strong multilingual capabilities with competitive performance on reasoning and coding benchmarks. Available for self-hosting and through various inference providers.

DeepSeek V3

DeepSeek's third-generation large language model featuring mixture-of-experts architecture, strong multilingual capabilities, and competitive performance on reasoning and coding benchmarks.

Command R7B

Cohere's compact 7B parameter model optimized for RAG, tool use, and code tasks. Delivers top-tier speed and efficiency on commodity GPUs and edge devices with 128K context window.

Phi-4

Microsoft's Phi-4 model with 14B parameters excelling at reasoning and code generation tasks, delivering strong performance relative to its compact size with efficient inference characteristics.

Cotype Nano

MTS AI's lightweight 1.5B parameter language model optimized for resource-constrained environments. Excels at Russian and English language tasks including content creation, translation, and text analysis. Runs efficiently on both CPU and GPU, including laptops and smartphones.

Aya Expanse 32B

Highly performant 32B multilingual language model from Cohere For AI, designed to rival monolingual model performance across 23 languages. Built using innovations in multilingual data arbitrage, direct preference optimization, and model merging techniques. Outperforms previous multilingual models on both automatic and human evaluations.

Llama 3.2 90B Vision Instruct

Meta's largest multimodal open-weight model with 90 billion parameters from the Llama 3.2 family. Delivers strong performance on both text and image understanding tasks with competitive results on visual reasoning benchmarks. Designed for high-quality inference requiring vision capabilities.

Llama 3.2 3B Instruct

Meta's lightweight open-weight model with 3 billion parameters from the Llama 3.2 family. Designed for on-device and edge deployment with strong text generation capabilities relative to its size. Supports instruction following and general-purpose tasks.

Llama 3.2 11B Vision Instruct

Meta's multimodal open-weight model with 11 billion parameters from the Llama 3.2 family. Supports both text and image inputs, enabling visual understanding tasks alongside standard text generation. Suitable for applications requiring vision capabilities at moderate scale.

Llama 3.1 8B Instruct

Meta's efficient open-weight model with 8 billion parameters from the Llama 3.1 family. Optimized for instruction following with strong performance on general tasks, coding, and multilingual benchmarks. Ideal for cost-effective deployment and edge inference scenarios.