Many users praise Grok 4.5 for hitting the Pareto frontier with strong coding performance at much lower cost, while others doubt it reaches GPT-5.5-level capability or criticize benchmark charts.
Based on 16 visible X reactions from 60 accounts; directional sample.
Ask a question below.
Published answers will appear here.
@rasbt Thanks for this update! Not sure why, but Grok Build + Grok 4.5 High is hugely underrated by the AI intelligentsia. It's a beast in terms of speed, token efficiency, and output/ outcomes. And a SOTA linter. https://x.com/alg0agent/status/2076066747946275136/photo/1
@grok @simone_m_romeo @rasbt Grok (you) hasn't received the Grok (model) upgrade, as evidenced by your failure to produce the correct Pareto-frontier chart, and again by your failure to understand humor:(
@rasbt The Pareto frontier observation is so right. You're getting frontier-level performance at one-fifth the cost from a week ago. That fundamentally changes what's possible to build.
@rasbt This image shows that - GPT 5.6 Sol ( xHigh & max) and - Opus 4.8 ( Max ) are completely waste of money and terrible in Cost effectiveness. GPT-5.6 Luna IS GREAT!
@rasbt Thanks for this update! Not sure why, but Grok Build + Grok 4.5 High is hugely underrated by the AI intelligentsia. It's a beast in terms of speed, token efficiency, and output/ outcomes. And a SOTA linter. https://x.com/alg0agent/status/2076066747946275136/photo/1
@grok @simone_m_romeo @rasbt Grok (you) hasn't received the Grok (model) upgrade, as evidenced by your failure to produce the correct Pareto-frontier chart, and again by your failure to understand humor:(
@rasbt The Pareto frontier observation is so right. You're getting frontier-level performance at one-fifth the cost from a week ago. That fundamentally changes what's possible to build.
@rasbt This image shows that - GPT 5.6 Sol ( xHigh & max) and - Opus 4.8 ( Max ) are completely waste of money and terrible in Cost effectiveness. GPT-5.6 Luna IS GREAT!
@rasbt Wait, you're telling me grok 4.5 is even better than fable, FABLE
@rasbt this setup actually looks super clean
Upon request, here's an updated version with Grok 4.5 and Meta's Muse Spark 1.1. Grok 4.5 seems to sit at the Pareto frontier. Good bang for the buck. (Also added harness info). https://x.com/rasbt/status/2075982283509571666/photo/1
Gemini where are you?? Whats the point of spark? How do you even use it? Is there a codex or Claude code equivalent? https://twitter.com/rasbt/status/2075982283509571666
@rasbt Where is the best open weight model at?
Many users praise Grok 4.5 for hitting the Pareto frontier with strong coding performance at much lower cost, while others doubt it reaches GPT-5.5-level capability or criticize benchmark charts.
Based on 16 visible X reactions from 60 accounts; directional sample.
Ask a question below.
Published answers will appear here.
@rasbt this setup actually looks super clean
Upon request, here's an updated version with Grok 4.5 and Meta's Muse Spark 1.1. Grok 4.5 seems to sit at the Pareto frontier. Good bang for the buck. (Also added harness info). https://x.com/rasbt/status/2075982283509571666/photo/1
Gemini where are you?? Whats the point of spark? How do you even use it? Is there a codex or Claude code equivalent? https://twitter.com/rasbt/status/2075982283509571666