Sarvam AI

Sarvam 30B is a 30B parameter Mixture-of-Experts chat and reasoning model from Sarvam AI, optimized for Indian languages, real-time conversation, high-throughput voice-agent pipelines, coding, and practical deployment. Sarvam documents 2.4B active parameters per token, 16T tokens of pre-training data, a 64K context window, Grouped Query Attention, Apache 2.0 open weights, and OpenAI-compatible chat completions.

Sarvam AI

Sarvam 105B is Sarvam AI's flagship 105B+ parameter Mixture-of-Experts reasoning model for Indian-language and English chat, complex reasoning, coding, long-context document analysis, and agentic tool-use workflows. Sarvam documents it as a 128K-context OpenAI-compatible chat model with Multi-head Latent Attention, 12T tokens of pre-training data, Apache 2.0 open weights, and production use powering Indus. Its strongest fit is Indian-language enterprise assistants, multilingual reasoning, and agent workflows where native script, romanized, and code-mixed inputs matter.

Anthropic

Anthropic's highest-capability widely released Claude model, documented for deep reasoning, codebase-scale work, long-context enterprise workloads, and multimodal inputs.

OpenAI

OpenAI's current flagship model for complex reasoning, coding, and professional work, documented in the OpenAI API model guide as the default starting point for high-complexity workloads.

Google

Google's advanced Gemini model for complex tasks, with official Gemini API documentation calling out deep reasoning and coding capabilities.

Anthropic

Anthropic's current Opus-tier Claude model, documented for complex reasoning, coding, and multimodal enterprise workloads below the newer Fable tier.

OpenAI

OpenAI's GPT-5.4 model, documented in the official OpenAI API model guide as part of the current GPT-5 family below the GPT-5.5 flagship lane.

Stability AI

Stable Diffusion XL (SDXL) 1.0 is Stability AI's latent diffusion text-to-image model for native 1024x1024 generation. The base model can run standalone or feed an optional refiner for the final denoising steps, and the published weights support self-hosted Diffusers workflows.

Alibaba

Qwen 2.5 72B Instruct is a large multilingual open-weights model from Alibaba’s Qwen family with strong coding and general chat performance. Common in APAC deployments and on Hugging Face inference endpoints—check license terms for commercial use.

Meta

Meta’s largest open-weights instruct checkpoint in the Llama 3.1 family, aimed at strong reasoning and coding quality with a permissive license for research and customization. It is typically served on dedicated GPU clusters or via partners (cloud inference, on-prem) rather than a single vendor API.

DeepSeek

DeepSeek-R1 is a reasoning-focused model family emphasizing chain-of-thought style behavior for math, code, and structured problem solving. Deployment options include API and open-weight variants—verify licensing and hosting constraints for your region.

Cohere

Cohere’s enterprise-oriented Command R+ emphasizes retrieval-grounded answers and tool orchestration patterns for business data. It targets teams building RAG-heavy assistants where citation-style behavior and connector patterns matter more than raw chat novelty.

Anthropic

Anthropic's Sonnet-tier model documented as the best combination of speed and intelligence in the Claude model overview.

AWS

Amazon Nova is AWS’s multimodal foundation model family for text, image, and video workloads delivered through Amazon Bedrock with enterprise IAM, VPC, and governance patterns. Model IDs and modalities vary by region—verify Bedrock model access lists.

Anysphere

Cursor's Composer 2.5 model, documented in the official Cursor changelog as a coding model used in the Auto routing flow.

Microsoft AI

Microsoft AI's agentic coding model in the MAI family, announced for fast code editing, debugging, and tool-driven developer workflows.

Mistral AI

Mistral's documented latest frontier-class multimodal model optimized for agentic and coding use cases.

xAI

xAI's documented default for general chat workloads, described in xAI docs as the most intelligent and fastest Grok model for non-specialized use cases.

Google

Google's fastest and most budget-friendly multimodal model in the Gemini 2.5 family, according to the Gemini API model documentation.

Microsoft AI

Microsoft AI's MAI image model for generation, editing, and visual content workflows, announced as part of the June 2026 MAI model release.

Microsoft AI

Microsoft AI's voice generation model in the MAI family, announced for natural text-to-speech and voice experiences.

Microsoft AI

Microsoft AI's speech-to-text model in the MAI family, announced for fast, accurate transcription across product surfaces.

OpenAI

OpenAI's smaller GPT-5.4 mini model, documented in the official OpenAI API model guide for lower-latency or lower-cost GPT-5 family workloads.

Mistral AI

Mistral's open-weight general-purpose multimodal model listed in official Mistral model documentation.