Tech & AICan It Code?Is Opus 5.5 the Best AI Model Yet?

Is Opus 5.5 the Best AI Model Yet?

2.7× channel medianLong-form6:07GamingPublished Sep 23, 2026, 1:28 PM · 3w ago
Views
128,178
+5 views/hour
Views / hour now
5
measured 7h ago
Peak views / hour
5
7h ago
First 24 hours
—
not tracked from upload
Engagement
1.4%
likes + comments / views
Likes
1.5K
1.2% of views
Comments
223
0.17% of views
Est. ad revenue
$897–$2,051
lifetime, $7–16 RPM

Checked every 12 h while the video is this age · last check 7h ago · next in 269 min. YouTube itself updates public counts with a short delay.

Avg views / day
4.8K
since upload
vs channel median
2.7×
this video's views ÷ the channel's usual
Rank in channel
#4
of 20 tracked

Views over time

2 readings

Total views Views gained per hour

Description

Claude Opus 5.5 came out yesterday, and on paper it beats both Claude Fable 5.1 and GPT-6 Astra. So I gave it the exact same three tests I ran last time for GPT-6 Astra and Fable 5.1: a Blender scene of a clockmaker's workshop in an old tower, interactive snow physics in Godot, and a full winter hunting game with walking, skiing, birds and a shotgun. Round one is a draw, round two goes to Opus 5.5, and in round three it wins by far: a character modelled and rigged from scratch in Blender, its own ski animations, deep footprints and flocking birds. The costs are estimated API costs at standard token prices, measured from the session logs (the main agent plus its helpers): Opus 5.5 cost $453 for all three tests, Fable 5.1 cost $472 and GPT-6 Astra cost $391. On my own Max 20x plan, all three Opus runs used around 10% of the weekly limit. The previous episode with GPT-6 Astra and Fable 5.1: https://www.youtube.com/watch?v=cbfDG0Bqmvk CHAPTERS 00:00 Big news: Opus 5.5 00:39 The benchmarks 00:51 Same three tests as last time 01:04 Test 1: Blender clock workshop 01:57 Test 1: cost and verdict 02:20 Test 2: snow physics in Godot 03:44 Test 2: cost and verdict 04:03 Test 3: winter hunting game 04:50 Skis, birds and shooting 05:21 Test 3: verdict and cost 05:39 Final score and total cost 05:49 10% of a Max 20x week CREDITS Benchmark table: "Introducing Claude Opus 5.5", Anthropic, https://www.anthropic.com/claude-opus-5-5 (screen recording). GPT-6 Astra and Claude Fable 5.1 footage comes from the previous episode of this channel. Motion capture used in the Opus 5.5 game: CMU Graphics Lab Motion Capture Database, http://mocap.cs.cmu.edu/. The data used in this project was obtained from mocap.cs.cmu.edu. The database was created with funding from NSF EIA-0196217. BVH conversion by Bruce Hahne (cgspeed). Textures and HDRIs used in the Opus 5.5 game: Poly Haven (https://polyhaven.com/license) and ambientCG, CC0. Sound effects used in the Opus 5.5 game: Pixabay, Pixabay Content License. Pixabay and Carnegie Mellon University logos via Wikimedia Commons. Godot, Blender, Poly Haven, Pixabay, Carnegie Mellon University, Claude and OpenAI names and logos are trademarks of their respective owners.
Tags
AI codingcan it codeAI builds a gameAI game developmentGodotGodot 4AI coding challengeGPT vs ClaudeClaude vs GPTAI vs AILLM comparisonAI programmingcoding with AIvibe codinggame devindie game devartificial intelligenceAI experimentsGPT 5Claude OpusGLM

Statistics from the YouTube Data API, tracked by FameX. Views per hour, revenue and comparisons are FameX estimates.