{"slug": "show-hn-27b-open-model-to-grade-robot-data-48x-cheaper-27x-faster-vs-6-astra", "title": "Show HN: 27B open model to grade robot data 48x cheaper/27x faster vs. 6 Astra", "summary": "Josef Chen released RoboJev, a 27B-parameter open-weights model (JEV-27B-VL, Apache-2.0) that grades robot task success from video, claiming it is 48x cheaper and 28x faster than GPT-6 Astra. Grading 68,555 episodes from 7,641 public LeRobot eval datasets on Hugging Face — 501 hours of video — costs $13.16 of GPU time, versus the $637 GPT-6 Astra would charge, according to Chen. The model was fine-tuned with LoRA on 52,598 robot episodes, and Chen released the weights, all three evaluation sets, a LeRobot reward plugin and a live-verdict video overlay.", "body_md": "Josef Chen on X: \"RoboJev: people are watching thousands of videos to label task success.\nNow, you can grade robot data faster than you make it. 48x cheaper and 28x faster than 6 Astra.\n27B parameters, open weights.\"\n\nRoboJev: people are watching thousands of videos to label task success.\nNow, you can grade robot data faster than you make it. 48x cheaper and 28x faster than 6 Astra.\n27B parameters, open weights.\n\nRoboJev: people are watching thousands of videos to label task success.\nNow, you can grade robot data faster than you make it. 48x cheaper and 28x faster than 6 Astra.\n27B parameters, open weights.\n\n3/ Grading 68,555 episodes from 7,641 public LeRobot eval datasets on Hugging Face, 501 hours of video, takes $13.16 of GPU time.\nGPT-6 Astra would charge $637.\n\n5/ How it's built: JEV-27B-VL, fine-tuned with LoRA on 52,598 robot episodes.\nOpen: weights (Apache-2.0), all three evaluation sets, the LeRobot reward plugin and the live-verdict overlay from the video.\nDetails and demo:\n\nTop: the ring lands and RoboJev goes to 1.00.\nBottom: the ring sits on the peg for a while, then falls off, and the score drops from 0.82 to 0.04.\nTry it on your own robot video: huggingface.co/spaces/josefch…", "url": "https://wpnews.pro/news/show-hn-27b-open-model-to-grade-robot-data-48x-cheaper-27x-faster-vs-6-astra", "canonical_source": "https://twitter.com/josefchen/status/2113687649131336039", "published_at": "2026-10-24 01:32:24+00:00", "updated_at": "2026-10-24 01:41:21.373775+00:00", "lang": "en", "topics": ["robotics", "large-language-models", "ai-research", "ai-tools", "ai-agents"], "entities": ["Josef Chen", "RoboJev", "JEV-27B-VL", "GPT-6 Astra", "Hugging Face", "LeRobot"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/show-hn-27b-open-model-to-grade-robot-data-48x-cheaper-27x-faster-vs-6-astra", "markdown": "https://wpnews.pro/news/show-hn-27b-open-model-to-grade-robot-data-48x-cheaper-27x-faster-vs-6-astra.md", "text": "https://wpnews.pro/news/show-hn-27b-open-model-to-grade-robot-data-48x-cheaper-27x-faster-vs-6-astra.txt", "jsonld": "https://wpnews.pro/news/show-hn-27b-open-model-to-grade-robot-data-48x-cheaper-27x-faster-vs-6-astra.jsonld"}}