Coverage
Which AI models Renown tracks
101 models and answer surfaces, 36 of them on every account without asking. The rest are added during onboarding when your market needs them. Verified 2026-08-26.
Two things are worth separating. An answer surface is where your buyer actually asks, such as ChatGPT or Google AI Overviews. A model is what generates the answer underneath. Most tools report the first and ignore the second, which is why they miss the regional models that decide answers outside English.
Answer surfaces
Every account gets all of these. This is where the buying questions get asked.
| Model | Maker | Coverage |
|---|---|---|
| ChatGPT | OpenAI | Default |
| Google AI Overviews | Default | |
| Google AI Mode | Default | |
| Gemini | Default | |
| Claude | Anthropic | Default |
| Perplexity | Perplexity | Default |
| Microsoft Copilot | Microsoft | Default |
| Grok | xAI | Default |
| DeepSeek | DeepSeek | Default |
| Meta AI | Meta | Default |
United States and Canada
The models behind most English-language answers.
| Model | Maker | Coverage |
|---|---|---|
| GPT-5.4 | OpenAI | Default |
| GPT-5.4 mini | OpenAI | Default |
| GPT-5.1 | OpenAI | Default |
| GPT-5 | OpenAI | Default |
| GPT-4.1 | OpenAI | On request |
| GPT-4o | OpenAI | On request |
| o4 | OpenAI | On request |
| o3 | OpenAI | On request |
| Claude Opus 5 | Anthropic | Default |
| Claude Sonnet 5 | Anthropic | Default |
| Claude Haiku 4.5 | Anthropic | Default |
| Claude Opus 4.1 | Anthropic | On request |
| Claude Sonnet 4.5 | Anthropic | On request |
| Gemini 3 Pro | Default | |
| Gemini 3 Flash | Default | |
| Gemini 2.5 Pro | On request | |
| Gemini 2.5 Flash | On request | |
| Gemma 3 | On request | |
| Llama 4 Maverick | Meta | Default |
| Llama 4 Scout | Meta | On request |
| Llama 3.3 | Meta | On request |
| Grok 4 | xAI | Default |
| Grok 3 | xAI | On request |
| Sonar | Perplexity | Default |
| Sonar Reasoning | Perplexity | On request |
| Nova Premier | Amazon | On request |
| Nova Pro | Amazon | On request |
| Nova Lite | Amazon | On request |
| Phi-4 | Microsoft | On request |
| MAI-1 | Microsoft | On request |
| Command A | Cohere | On request |
| Command R+ | Cohere | On request |
| Nemotron | NVIDIA | On request |
| OLMo 2 | Allen Institute for AI | On request |
Europe
What answers look like in European languages, and to European buyers.
| Model | Maker | Coverage |
|---|---|---|
| Mistral Large 3 | Mistral AI, France | Default |
| Mistral Medium 3 | Mistral AI, France | On request |
| Magistral | Mistral AI, France | On request |
| Devstral | Mistral AI, France | On request |
| Le Chat | Mistral AI, France | Default |
| EuroLLM | EU consortium | On request |
| Teuken-7B | OpenGPT-X, Germany | On request |
| Salamandra | Barcelona Supercomputing Center, Spain | On request |
| Occiglot | European research consortium | On request |
| Jamba | AI21 Labs, Israel | On request |
Nordics and Finland
Finnish and Nordic answers are not the English answer translated. They are built from different sources, so they are measured separately.
| Model | Maker | Coverage |
|---|---|---|
| Poro 2 | Silo AI, TurkuNLP and HPLT, Finland | Default |
| Viking | Silo AI, TurkuNLP and HPLT, Finland | Default |
| GPT-SW3 | AI Sweden | On request |
| NorMistral | University of Oslo, Norway | On request |
| NB-GPT | National Library of Norway | On request |
| Danoliterate | Danish research | On request |
| Mistral Large 3, Nordic prompts | Mistral AI, France | Default |
Middle East and North Africa
Most tools in this category track none of the state-backed Arabic models. Renown treats Arabic as its own answer market, not a translation of the English one.
| Model | Maker | Coverage |
|---|---|---|
| Falcon-H1 | TII, United Arab Emirates | Default |
| Falcon 3 | TII, United Arab Emirates | On request |
| ALLaM | SDAIA, Saudi Arabia | Default |
| Jais 2 | G42 and Core42, United Arab Emirates | Default |
| Fanar | QCRI, Qatar | Default |
| Noor | TII, United Arab Emirates | On request |
| AceGPT | KAUST, Saudi Arabia | On request |
India
Indian-language answers draw on a different source set again, and are measured per language rather than as one market.
| Model | Maker | Coverage |
|---|---|---|
| Sarvam-M | Sarvam AI, India | Default |
| Sarvam 1 | Sarvam AI, India | On request |
| Krutrim 2 | Ola Krutrim, India | Default |
| BharatGPT | CoRover, India | On request |
| Hanooman | SML and BharatGPT group, India | On request |
| Project Indus | Tech Mahindra, India | On request |
| Airavata | AI4Bharat, India | On request |
| Nanda | MBZUAI and partners | On request |
China and East Asia
Relevant for brands selling into these markets, and for global models trained partly on their output.
| Model | Maker | Coverage |
|---|---|---|
| DeepSeek V3.2 | DeepSeek, China | Default |
| DeepSeek R1 | DeepSeek, China | On request |
| Qwen3-Max | Alibaba, China | Default |
| Qwen3 | Alibaba, China | On request |
| Kimi K2 | Moonshot AI, China | Default |
| GLM-4.6 | Zhipu AI, China | On request |
| MiniMax M2 | MiniMax, China | On request |
| Ernie 4.5 | Baidu, China | On request |
| Hunyuan | Tencent, China | On request |
| Doubao | ByteDance, China | On request |
| Yi Large | 01.AI, China | On request |
| Baichuan 4 | Baichuan, China | On request |
| HyperCLOVA X | Naver, South Korea | On request |
| Solar Pro | Upstage, South Korea | On request |
| PLaMo | Preferred Networks, Japan | On request |
| Swallow | Tokyo Institute of Technology, Japan | On request |
Southeast Asia, Africa and Latin America
Regional models that shape answers where the global assistants are weakest.
| Model | Maker | Coverage |
|---|---|---|
| SEA-LION | AI Singapore | On request |
| Sailor 2 | SEA regional consortium | On request |
| Typhoon 2 | SCB 10X, Thailand | On request |
| PhoGPT | VinAI, Vietnam | On request |
| Sahabat-AI | Indosat and GoTo, Indonesia | On request |
| InkubaLM | Lelapa AI, South Africa | On request |
| Ubuntu-LLM | African research consortium | On request |
| Latam-GPT | CENIA, Chile | On request |
| Sabiá 3 | Maritaca AI, Brazil | On request |
Languages
The 150+ most spoken languages are covered by default. That includes Finnish, Swedish, Norwegian, Danish, every major European language, Arabic, Hindi, Tamil, Bengali, Telugu, Marathi and the other major Indian languages. Anything outside that set is supported on request.
Language matters more than most teams expect. Our Helsinki fine dining study found that English and Finnish prompts return different restaurants for the same question, which means a brand can lead the tourist answer and be invisible to locals. One language is half a measurement.
How a model gets added
Ask. If a model shapes answers in your category or your market and it is marked on request above, it goes into your account during onboarding at no extra cost. We would rather measure the six models your buyers actually use than publish a number.
Missing something that is not on this page at all? Tell us and we will add it and update this list.