[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

[@LakshyAAAgrawal](/creator/twitter/LakshyAAAgrawal)
"Optimizing a data analysis coding agent with GEPA using execution-guided feedback on real-world workloads. Amazing tutorial by @ArslanSAAS:"  
[X Link](https://x.com/LakshyAAAgrawal/status/1977147206168658216) [@LakshyAAAgrawal](/creator/x/LakshyAAAgrawal) 2025-10-11T23:00Z 2722 followers, 10.3K engagements


"In this context GEPA works as a prompt optimizer so the end result is a prompt (or multiple prompts for a multi-agent system one for each component). However one aspect that does not get highlighted enough is that GEPA is a text evolution engine: Given a target metric GEPA can efficiently search/evolve the right text to improve that metric. What the text represents is upto the user. For example In this notebook we use text to represent a full agent code and GEPA ends up discovering a very sophisticated agent (that can perform self-reflection and iterative refinement on code) for ARC-AGI"  
[X Link](https://x.com/LakshyAAAgrawal/status/1968236513810087975) [@LakshyAAAgrawal](/creator/x/LakshyAAAgrawal) 2025-09-17T08:52Z 2720 followers, 104.2K engagements


"Struggling to use LLMs for creative tasks @hammer_mt talks about the powerful "Evaluator-Optimizer" pattern with GEPA+@DSPyOSS to optimize prompts for fuzzy generative tasks where evals are informal and subjective. Checkout the full talk prompt and executable notebook below"  
[X Link](https://x.com/LakshyAAAgrawal/status/1979389591389114796) [@LakshyAAAgrawal](/creator/x/LakshyAAAgrawal) 2025-10-18T03:30Z 2721 followers, 8267 engagements


"Super cool work inspired by @karpathy's tweet about converting textbook practice problems into environments. What a cool topic to build environment around too"  
[X Link](https://x.com/LakshyAAAgrawal/status/1979652060539990255) [@LakshyAAAgrawal](/creator/x/LakshyAAAgrawal) 2025-10-18T20:53Z 2721 followers, 3464 engagements


"Checkout the GEPA repository for optimizing your agents:"  
[X Link](https://x.com/LakshyAAAgrawal/status/1979664893885530204) [@LakshyAAAgrawal](/creator/x/LakshyAAAgrawal) 2025-10-18T21:44Z 2722 followers, XXX engagements


"Check the list at:"  
[X Link](https://x.com/LakshyAAAgrawal/status/1979664895194100072) [@LakshyAAAgrawal](/creator/x/LakshyAAAgrawal) 2025-10-18T21:44Z 2720 followers, XXX engagements


"Hi @VictorTaelin Big fan of your work I would be super excited if you could try GEPA for this task (as I am very excited about the tasks you are doing). I would be happy to jump on a call to understand and discuss what might be the best way to integrate GEPA into your pipeline. The nice side-effect as highlighted by @gooby_esq is that once you have such a setup you wouldn't necessarily be tied to GEPA but could then also explore RL / SFT among other techniques if you want. In a nutshell for GEPA optimization you would have to provide a list of example inputs ("implement a function X to do Y")"  
[X Link](https://x.com/LakshyAAAgrawal/status/1979014429653262750) [@LakshyAAAgrawal](/creator/x/LakshyAAAgrawal) 2025-10-17T02:39Z 2709 followers, XX engagements


"@2abstract4me @casper_hansen_ Hey you can use an LLM-judge even without labels to perform GEPA optimization. Example at"  
[X Link](https://x.com/LakshyAAAgrawal/status/1979048007396979138) [@LakshyAAAgrawal](/creator/x/LakshyAAAgrawal) 2025-10-17T04:53Z 2709 followers, XX engagements


"@elonmusk GEPA can be very useful tool to help users migrate their existing LLM-based workflows and agents to new models across model families"  
[X Link](https://x.com/LakshyAAAgrawal/status/1976355709890723999) [@LakshyAAAgrawal](/creator/x/LakshyAAAgrawal) 2025-10-09T18:34Z 2719 followers, 1766 engagements


"@elonmusk Databricks achieved 90x cost savings with GEPA:"  
[X Link](https://x.com/LakshyAAAgrawal/status/1976356538320290142) [@LakshyAAAgrawal](/creator/x/LakshyAAAgrawal) 2025-10-09T18:38Z 2719 followers, XXX engagements


"@madhavsinghal_ Thanks it has been validated with models at all scales (on the small end with Gemma X to Claude Opus Gemini-2.5 Pro on the large end). Some examples below"  
[X Link](https://x.com/LakshyAAAgrawal/status/1979792101765521416) [@LakshyAAAgrawal](/creator/x/LakshyAAAgrawal) 2025-10-19T06:09Z 2718 followers, XX engagements


"@madhavsinghal_ Databricks' highlights large gains on Claude Opus XXX Sonnet X and GPT-OSS 120B:"  
[X Link](https://x.com/LakshyAAAgrawal/status/1979792124032978966) [@LakshyAAAgrawal](/creator/x/LakshyAAAgrawal) 2025-10-19T06:09Z 2720 followers, XX engagements