CS2 update · Community Announcements ·

Animgraph 2 Beta

Valve published "Animgraph 2 Beta" on 1 April 2026 in Community Announcements. The PatinaAI market index's median read 32.24 points in the seven days after against 32.15 before, +0.3 %. A typical week moves under 0.1 % either way on the same measure. A read is a comparison with a date, not a cause.

+0.3 %index, 7d after vs before
32.15median before, points
32.24median after, points
The postCommunity Announcements ·

https://clan.akamai.steamstatic.com/images/3381077/3b774ae067b0a4fb8b71d41fdde8d073ac5a064c.png

Today we're shipping a beta build that updates CS2's animation system to Animgraph 2, which reduces the CPU and networking costs associated with animation. Along with the technical work involved in switching to the new system, all third person animations have been re-authored and, in many cases, adjusted in response to player feedback.

It's time for you to take a look, poke around for bugs, and tell us what you think. To opt into the beta build, follow the instructions here and choose the 'animgraph_2_beta' build. Note that you will not be able to connect to Valve servers while you are using the beta build. To report bugs or provide feedback about Animgraph 2, please email csgoteamfeedback@valvesoftware.com with the subject "AG2 Beta".

Read the post on Steam. Text as Steam's news feed serves it.

Market afterthe days after the post, the venues' daily prices

No read: the venues' daily prices do not cover the days around this post.

Index read to

The PatinaAI market index's median over the seven days after the post was 32.24 points against 32.15 over the seven days before: +0.3 %. A typical week moves under 0.1 % either way on the same measure, the drift this read has to beat. The post's own day is left out. A comparison with a date, not causation.

Mentions1 named

The item types and weapons the post names, in the order it names them. A word match, not a reading of the post.

How to read the number

The read compares the market index's median over the seven UTC days after the post with the seven days before it, the post's own day left out, and needs at least three published days on each side. The same measure taken on every day of the series is the drift a typical week shows; a read inside that drift is the market doing what it was doing. Neither is proof the post moved anything. The composite is on the market report, every basket under the indexes, and every update is on one list.