🐦 Twitter Post Details

Viewing enriched Twitter post

@LiorOnAI

Most language models only read forward. Perplexity just open-sourced 4 models that read text in both directions. They used a technique from image generation to retrain Qwen3 so every word can see every other word in a passage. That changes how well a model understands meaning. They built four models from this: 1. Two sizes: 0.6B and 4B parameters 2. Two types: standard search embeddings and context-aware embeddings The context-aware version is the interesting one. It processes an entire document at once, so each small chunk "knows" what the full document is about. Standard embeddings treat each chunk in isolation. > Tops benchmarks for models of similar size > Works in multiple languages out of the box > MIT licensed, free for commercial use If you're building search over large document collections, you can now get document-level understanding without running a massive model. Small enough to actually deploy.

Media 1

📊 Media Metadata

{
  "media": [
    {
      "type": "photo",
      "url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2027483180752900129/media_0.jpg?",
      "filename": "media_0.jpg"
    }
  ],
  "processed_at": "2026-03-01T19:17:17.935081",
  "pipeline_version": "2.0"
}

🔧 Raw API Response

{
  "type": "tweet",
  "id": "2027483180752900129",
  "url": "https://x.com/LiorOnAI/status/2027483180752900129",
  "twitterUrl": "https://twitter.com/LiorOnAI/status/2027483180752900129",
  "text": "Most language models only read forward.\n\nPerplexity just open-sourced 4 models that read text in both directions.\n\nThey used a technique from image generation to retrain Qwen3 so every word can see every other word in a passage. \n\nThat changes how well a model understands meaning.\n\nThey built four models from this:\n\n1. Two sizes: 0.6B and 4B parameters\n\n2. Two types: standard search embeddings and context-aware embeddings\n\nThe context-aware version is the interesting one. \n\nIt processes an entire document at once, so each small chunk \"knows\" what the full document is about. Standard embeddings treat each chunk in isolation.\n\n> Tops benchmarks for models of similar size\n> Works in multiple languages out of the box\n> MIT licensed, free for commercial use\n\nIf you're building search over large document collections, you can now get document-level understanding without running a massive model. \n\nSmall enough to actually deploy.",
  "source": "Twitter for iPhone",
  "retweetCount": 10,
  "replyCount": 5,
  "likeCount": 40,
  "quoteCount": 0,
  "viewCount": 3884,
  "createdAt": "Fri Feb 27 20:37:13 +0000 2026",
  "lang": "en",
  "bookmarkCount": 39,
  "isReply": false,
  "inReplyToId": null,
  "conversationId": "2027483180752900129",
  "displayTextRange": [
    0,
    272
  ],
  "inReplyToUserId": null,
  "inReplyToUsername": null,
  "author": {
    "type": "user",
    "userName": "LiorOnAI",
    "url": "https://x.com/LiorOnAI",
    "twitterUrl": "https://twitter.com/LiorOnAI",
    "id": "931470139",
    "name": "Lior Alexander",
    "isVerified": false,
    "isBlueVerified": true,
    "verifiedType": null,
    "profilePicture": "https://pbs.twimg.com/profile_images/2027106343283527680/lh729xEs_normal.jpg",
    "coverPicture": "https://pbs.twimg.com/profile_banners/931470139/1761077189",
    "description": "",
    "location": "",
    "followers": 112932,
    "following": 2153,
    "status": "",
    "canDm": true,
    "canMediaTag": false,
    "createdAt": "Wed Nov 07 07:19:36 +0000 2012",
    "entities": {
      "description": {
        "urls": []
      },
      "url": {}
    },
    "fastFollowersCount": 0,
    "favouritesCount": 6770,
    "hasCustomTimelines": true,
    "isTranslator": false,
    "mediaCount": 661,
    "statusesCount": 3756,
    "withheldInCountries": [],
    "affiliatesHighlightedLabel": {},
    "possiblySensitive": false,
    "pinnedTweetIds": [],
    "profile_bio": {
      "description": "Covering the latest news for AI devs • Founder @AlphaSignalAI (270k users) •  ML Eng since 2017 • Ex-Mila • MIT",
      "entities": {
        "description": {
          "hashtags": [],
          "symbols": [],
          "urls": [],
          "user_mentions": [
            {
              "id_str": "0",
              "indices": [
                47,
                61
              ],
              "name": "",
              "screen_name": "AlphaSignalAI"
            }
          ]
        },
        "url": {
          "urls": [
            {
              "display_url": "alphasignal.ai",
              "expanded_url": "https://alphasignal.ai",
              "indices": [
                0,
                23
              ],
              "url": "https://t.co/AyubevadmD"
            }
          ]
        }
      }
    },
    "isAutomated": false,
    "automatedBy": null
  },
  "extendedEntities": {
    "media": [
      {
        "allow_download_status": {
          "allow_download": true
        },
        "display_url": "pic.twitter.com/VvcZ3l7Xa4",
        "expanded_url": "https://twitter.com/LiorOnAI/status/2027483180752900129/photo/1",
        "ext_media_availability": {
          "status": "Available"
        },
        "features": {
          "large": {
            "faces": [
              {
                "h": 104,
                "w": 104,
                "x": 746,
                "y": 478
              }
            ]
          },
          "orig": {
            "faces": [
              {
                "h": 104,
                "w": 104,
                "x": 746,
                "y": 478
              }
            ]
          }
        },
        "id_str": "2027481912659636224",
        "indices": [
          273,
          296
        ],
        "media_key": "3_2027481912659636224",
        "media_results": {
          "id": "QXBpTWVkaWFSZXN1bHRzOgwAAQoAARwjEA64m3AACgACHCMRNfjbECEAAA==",
          "result": {
            "__typename": "ApiMedia",
            "id": "QXBpTWVkaWE6DAABCgABHCMQDribcAAKAAIcIxE1+NsQIQAA",
            "media_key": "3_2027481912659636224"
          }
        },
        "media_url_https": "https://pbs.twimg.com/media/HCMQDribcAASgVA.jpg",
        "original_info": {
          "focus_rects": [
            {
              "h": 1127,
              "w": 2013,
              "x": 2,
              "y": 0
            },
            {
              "h": 1127,
              "w": 1127,
              "x": 445,
              "y": 0
            },
            {
              "h": 1127,
              "w": 989,
              "x": 514,
              "y": 0
            },
            {
              "h": 1127,
              "w": 564,
              "x": 726,
              "y": 0
            },
            {
              "h": 1127,
              "w": 2016,
              "x": 0,
              "y": 0
            }
          ],
          "height": 1127,
          "width": 2016
        },
        "sizes": {
          "large": {
            "h": 1127,
            "w": 2016
          }
        },
        "type": "photo",
        "url": "https://t.co/VvcZ3l7Xa4"
      }
    ]
  },
  "card": null,
  "place": {},
  "entities": {
    "hashtags": [],
    "symbols": [],
    "urls": [],
    "user_mentions": []
  },
  "quoted_tweet": null,
  "retweeted_tweet": null,
  "isLimitedReply": false,
  "article": null
}