Perplexity's tokenmaxxing Model Council gives you multiple bot perspectives

Perplexity has expanded its Model Council to its cloud-based Computer platform, providing users with more options for configuring multi-model queries. The concept behind this Model Council is to enable users to create a board of top models for a single issue, with each model tackling the problem independently and generating a synthesis report.

Perplexity's Model Council functions as a skill that instructs an orchestrator model to spin up parallel subagents, each acting as a full agent and addressing the issue independently. Once every perspective is complete, the chosen model, acting as the 'chair,' synthesizes the outputs, highlighting both the points of agreement and divergence.

Initially, the Model Council was limited to three models, but the recent update now allows users to select between two and eight models, including those from frontier labs like OpenAI, Anthropic, and Google, as well as open-weight models such as GLM and Kimi. Computer, which is essentially a web interface for a model orchestration layer, can now present the synthesis as various work-ready assets, such as reports or board decks.

This expanded Model Council is particularly useful for ambiguous questions involving judgment calls, trade-offs, risk assessment, or other situations where there's no single right answer, as model-to-model dialogue highlights consensus points and conflict areas. The Model Council can be employed for a variety of use cases, including legal queries, financial questions, corporate decision-making, business growth modeling, and engineering.

Perplexity also offers users the option to inquire about confidence levels, allowing them to assess the collective answer's reliability. However, this enhanced Model Council comes at a cost, as access was previously restricted to enterprise tiers and individual Max customers. With the addition of Computer, Model Council is now available to Pro ($20/month), Max ($200/month), and individual Max users, as well as enterprise tiers.

Users on Pro receive 4,000 bonus credits, while Max customers get 35,000 and an additional 10,000 monthly. Enterprise Pro users start with 8,000 bonus credits and receive 500 per seat monthly, while Enterprise Max users begin with a 45,000 credit pool and get 15,000 per seat monthly. The pricing structure for Computer credits, which function at the product level, is based on a $1 value for 100 credits.

A "light" task, like summarizing a document, can consume between 100 and 350 credits, while a "heavy" task, such as creating work-ready assets, can require up to 2,275 credits. The cost of using the Model Council depends on the complexity of the query, with simpler tasks costing less than a penny and more complex tasks being significantly more expensive.