A fast assistant, research tool, and a coding agent don’t all need the same model. This latest group of additions gives you more ways to choose a model that fits the work: GPT-6.1 Terra, Mistral Small 4.0, Mistral Large 4.0, DeepSeek V4.1 Flash 1017, and Gemini 4 Argon!
OpenAI
- GPT-6.1 Terra is an efficient option for reasoning, writing, and focused agent tasks! It accepts text and images and offers configurable reasoning, making it worth testing when you want a responsive agent that can still work through visual inputs.
Mistral
- Mistral Small 4.0 combines instruction following, reasoning, coding, and multimodal input in a 256K context window. It’s a useful candidate for everyday agents that need to handle varied requests efficiently.
- Mistral Large 4.0 is Mistral’s multimodal flagship in public preview. With optional reasoning, tool calling, and a one-million-token context window, it’s aimed at more demanding tasks involving substantial context.
DeepSeek
- DeepSeek V4.1 Flash 1017 is the October 17 snapshot of V4.1 Flash. It brings native image understanding for coding, visual analysis, and long-context agent workflows.
- Gemini 4 Argon is Google’s flagship multimodal reasoning option for complex coding, scientific problem-solving, and autonomous agent workflows.
You can find all five in the builder model selector! Try them out on your agents and let us know if you notice any differences in responses, speed, and cost!