Omnimatch Supported models
Cloud and on-premises share the same Omnimatch agent model catalogue. Models are grouped below by their provider. Tool-code models are also listed, with their use shown separately.
ZDR capable means the model supports zero data retention: the provider can process your requests without storing your prompts or responses.
| Model | Available for | Context (tokens) | ZDR capable |
|---|---|---|---|
| Anthropic | |||
| Claude Fable 5.1 | Agent chat | 1,000,000 | Yes |
| Claude Haiku 4.5 | Agent chat | 200,000 | Yes |
| Claude Opus 5 | Agent chat | 200,000 | Yes |
| Claude Sonnet 5 | Agent chat | 200,000 | Yes |
| DeepSeek | |||
| DeepSeek R1 | Agent chat | 163,840 | No |
| DeepSeek R1 Distill 32B | Tool code | 80,000 | No |
| DeepSeek V3.2 | Agent chat | 163,840 | No |
| Gemini 3.1 Flash Lite | Agent chat | 1,048,576 | No |
| Gemini 3.1 Pro | Agent chat | 1,048,576 | No |
| Gemini 3.6 Flash | Agent chat | 1,048,576 | No |
| Gemma 4 26B | Agent chat | 256,000 | No |
| Gemma 4 26B No Thinking | Agent chat | 256,000 | No |
| IBM | |||
| Granite 4.0 Micro | Agent chat | 131,000 | No |
| Meta | |||
| Llama 4 Scout | Agent chat | 131,000 | No |
| Mistral | |||
| Mistral Large | Agent chat | 131,072 | No |
| Mistral Small 3.1 | Agent chat | 128,000 | No |
| Moonshot AI | |||
| Kimi K2.6 | Agent chat | 262,144 | No |
| Kimi K2.6 No Thinking | Agent chat | 262,144 | No |
| Kimi K2.7 Code | Agent chat, Tool code | 262,144 | No |
| NVIDIA | |||
| Nemotron 3 120B | Agent chat | 256,000 | No |
| OpenAI | |||
| GPT-5.4 Mini | Agent chat | 400,000 | Yes |
| GPT-5.4 Nano | Agent chat | 400,000 | Yes |
| GPT-5.6 | Agent chat | 400,000 | Yes |
| GPT-5.6 Fast | Agent chat | 400,000 | Yes |
| GPT-5.6 Luna | Agent chat | 400,000 | Yes |
| GPT-5.6 Terra | Agent chat | 400,000 | Yes |
| GPT-6 Astra | Agent chat | 1,050,000 | Yes |
| GPT-OSS 120B | Agent chat, Tool code | 128,000 | No |
| GPT-OSS 20B | Agent chat | 128,000 | No |
| Qwen | |||
| Qwen 3.8 27B | Agent chat | 262,144 | No |
| Qwen 3.8 27B No Thinking | Agent chat | 262,144 | No |
| Qwen3 30B | Tool code | 32,768 | No |
| Z.ai | |||
| GLM 4.7 Flash | Agent chat | 131,072 | No |
| GLM 4.7 Flash No Thinking | Agent chat | 131,072 | No |
| GLM 5.2 | Agent chat | 262,144 | No |
| GLM 5.2 No Thinking | Agent chat | 262,144 | No |
| GLM 5.3 | Agent chat | 1,310,720 | No |
| GLM 5.3 Flash | Agent chat | 1,048,576 | No |
| xAI | |||
| Grok 4.1 Fast | Agent chat | 2,000,000 | No |
| Grok 4.3 | Agent chat | 256,000 | No |
| Grok 4.5 | Agent chat | 256,000 | No |
For setup, see Choose your agent’s LLM or on-premises model settings. On-premises access to Omnimatch models requires internet access, a licence with model access and Omnimatch credit. Your own local or external provider can offer additional models; its retention terms depend on your configuration and agreement.