by UnbiasedReleased 1 October 2026
A preview of the next Pareto, a multimodal composite model for research, coding and agentic work that reads text and images across a million-token window and calls tools, and may change without notice
Reads about 1 000 pages · writes up to 130 pages
Standard OpenAI fields, grouped by what they control. Anything not listed is accepted and ignored.
Shaping the reply
How long, how varied, where to stop.
max_tokenstemperaturetop_pTools
Let the model call your functions.
toolstool_choiceNo. Unbiased has retired it. Other Unbiased models are on sale.
A preview of the next Pareto, a multimodal composite model for research, coding and agentic work that reads text and images across a million-token window and calls tools, and may change without notice.
Unbiased names no languages for Pareto 26.10 Preview. Arabic, French and Darja: not stated.
It reads text and images. Up to 1M tokens in one request, about 1 000 pages. It writes up to 98 000 words at once.
No. Unbiased has not published its weights.
Unbiased. Released 1 October 2026.
Pareto 26.10 is a multimodal composite model for research, coding and agentic workflows that reads text and images, calls your tools and holds a million tokens in view, with frontier-level performance on general-purpose tasks
A multimodal composite model for research, coding and agentic work that runs several models on each request without switching mid-conversation, reads text and images in a 262,144-token window and calls your tools
From what its maker and two benchmark sites publish, and where it stands on OpenDunes. Read 15 Oct 2026.
Thinks first No.
Works through the problem before it answers.
Uses your tools
Can call functions your code provides.
Fills your JSON No.
Answers in the JSON shape you give it.
Searches the web No.
Can look things up online while it answers.
Cheaper on repeats
A prompt it saw a moment ago costs less the second time.
Reads screenshots and documents alongside code to drive agentic coding steps
Uses your tools
Keeps long research sessions in context across a million tokens of mixed input
Works through command-line tasks step by step, calling your tools to run each command
Uses your tools