Kat’s verdictUpgrade beyond level 100 only on a dragon already doing a defined combat job, and only where the next displayed effect helps that job. Talent 1 and Talent 2 work together when the dragon is deployed. Talent 3 is a separate active skill with a fragment cost, so do not mix it into a permanent-talent comparison.

First, correct the ceiling

Evony’s official version 5.17.0 notes, published 2026-01-16, raised the maximum level of Dragon Talent 1 and Talent 2 to 120. The same notice says the Active Talent cap did not change. An official Q&A updated 2025-12-23 explains the operating rule: Talent 1 and Talent 2 are active together with no switching when the dragon is deployed in battle; Talent 3 must be activated separately and consumes a specified number of Dragon Source Fragments.

This matters because guides written before 2026-01-16 can still show 100 as the maximum. One detailed guide dated 2026-01-12 does exactly that. Its old ceiling is useful history, not a current stopping rule. Check the publication date before copying any talent table. Dragons age slowly. Search results are more adventurous.

Give the dragon one current job

A specialist guide dated 2026-09-21 also recommends concentrating development on dragons assigned to the strongest generals instead of spreading materials across every dragon. That is sensible prioritisation, not an official combat formula. The in-game condition text remains the deciding evidence.

Audit levels 101 to 120 from the live screen

Before the first upgrade, record the current talent level, current effect, next-level effect and every quoted material. For each relevant line, calculate: marginal useful gain = next-level effect − current effect. Then compare the quoted Gems, Dragon Crystals or Dragon Source Fragments with the next permanent upgrade available to the same march.

Do not multiply one early level by 20 and call it the total. The official notice confirms the new cap, not a flat cost curve or a flat gain curve. If the client shows different prices or gains at later levels, the live screen wins. Save a screenshot at level 100 and another at each stopping point. A dragon is expensive enough without memory doing the accounting.

Keep permanent talents and Active Talent apart

Talent 1 and Talent 2 are simultaneous passive effects when the dragon is deployed. Talent 3 is selected in the Active Skill interface and consumes Dragon Source Fragments when activated. The version 5.17.0 cap change therefore does not prove that Talent 3 became stronger, cheaper or more available.

Use two ledgers. Put levels 101 to 120 of Talent 1 and Talent 2 in the permanent-build ledger. Put each planned Talent 3 activation in the consumable-event ledger. If the same fragments compete for both, compare the next permanent gain with the number of active uses you are giving up. “Dragon materials” is not one budget merely because the icons share a room.

Use five stopping rules

Evidence limits

Official material confirms the 120 cap for Talent 1 and Talent 2, the unchanged Active Talent cap, simultaneous use of Talent 1 and Talent 2, and separate activation of Talent 3 with Dragon Source Fragments. It does not publish a complete level-by-level cost table, a universal upgrade order or a promise that every dragon and role receives equal value. Role assignments, assistant behaviour and investment order are specialist guidance. Current in-game text, prices and battle conditions take priority.

Sources: official version 5.17.0 Dragon Talent change, published 2026-01-16; official Dragon Talent activation Q&A, updated 2025-12-23; specialist Dragon Talent and material guide; specialist dragon role and investment guide, updated 2026-07-21; specialist dragon role guide, published 2026-09-21; pre-change specialist guide showing the former 100 cap, published 2026-01-12.

Evidence reviewed 2026-10-21. Official rules, dated specialist guidance and editorial decision tests are separated. No current package price, flat 101-to-120 cost curve, assistant buff test or guaranteed battle result is claimed.