🐦 Twitter Post Details

Viewing enriched Twitter post

@vllm_project

We’ve published an official vLLM Recipe for serving XiaomiMiMo/MiMo-V2-Flash—including tool calling, DP/TP/EP configs, and the key knobs to tune context length, latency, and KV cache. Commands are in the images; highlights below. 💡 Tips: - Set --max-model-len to manage memory (common: 65536, max: 128k) - Balance throughput vs latency with --max-num-batched-tokens (prompt-heavy: 32768; lower to 16k/8k for lower latency/activation mem) - Increase KV cache with --gpu-memory-utilization 0.95 (default is conservative) - Tool calling needs the right parsers (see screenshot flags) - “Thinking mode” in API: set "enable_thinking": true; disable by setting it to false or removing chat_template_kwargs 📑 Recipe + full details: https://t.co/MpUFWCPGjm Thanks to @XiaomiMiMo and the community for the contributions!

Media 1

📊 Media Metadata

{
  "media": [
    {
      "url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2002938138549682366/media_0.jpg?",
      "media_url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/2002938138549682366/media_0.jpg?",
      "type": "photo",
      "filename": "media_0.jpg"
    }
  ],
  "processed_at": "2025-12-22T06:19:57.515548",
  "pipeline_version": "2.0"
}

🔧 Raw API Response

{
  "type": "tweet",
  "id": "2002938138549682366",
  "url": "https://x.com/vllm_project/status/2002938138549682366",
  "twitterUrl": "https://twitter.com/vllm_project/status/2002938138549682366",
  "text": "We’ve published an official vLLM Recipe for serving XiaomiMiMo/MiMo-V2-Flash—including tool calling, DP/TP/EP configs, and the key knobs to tune context length, latency, and KV cache. Commands are in the images; highlights below.\n\n💡 Tips:\n  - Set --max-model-len to manage memory (common: 65536, max: 128k)\n  - Balance throughput vs latency with --max-num-batched-tokens (prompt-heavy: 32768; lower to 16k/8k for lower latency/activation mem)\n  - Increase KV cache with --gpu-memory-utilization 0.95 (default is conservative)\n  - Tool calling needs the right parsers (see screenshot flags)\n  - “Thinking mode” in API: set \"enable_thinking\": true; disable by setting it to false or removing chat_template_kwargs\n\n📑 Recipe + full details: https://t.co/MpUFWCPGjm\nThanks to @XiaomiMiMo and the community for the contributions!",
  "source": "Twitter for iPhone",
  "retweetCount": 5,
  "replyCount": 1,
  "likeCount": 70,
  "quoteCount": 0,
  "viewCount": 3287,
  "createdAt": "Mon Dec 22 03:03:59 +0000 2025",
  "lang": "en",
  "bookmarkCount": 17,
  "isReply": false,
  "inReplyToId": null,
  "conversationId": "2002938138549682366",
  "displayTextRange": [
    0,
    280
  ],
  "inReplyToUserId": null,
  "inReplyToUsername": null,
  "author": {
    "type": "user",
    "userName": "vllm_project",
    "url": "https://x.com/vllm_project",
    "twitterUrl": "https://twitter.com/vllm_project",
    "id": "1774187564276289536",
    "name": "vLLM",
    "isVerified": false,
    "isBlueVerified": true,
    "verifiedType": null,
    "profilePicture": "https://pbs.twimg.com/profile_images/1774187681746182144/N_5NJ8B1_normal.jpg",
    "coverPicture": "https://pbs.twimg.com/profile_banners/1774187564276289536/1733806062",
    "description": "A high-throughput and memory-efficient inference and serving engine for LLMs. Join https://t.co/lxJ0SfX5pJ to discuss together with the community!",
    "location": "",
    "followers": 27259,
    "following": 27,
    "status": "",
    "canDm": true,
    "canMediaTag": true,
    "createdAt": "Sat Mar 30 21:31:01 +0000 2024",
    "entities": {
      "description": {
        "urls": [
          {
            "display_url": "slack.vllm.ai",
            "expanded_url": "http://slack.vllm.ai",
            "url": "https://t.co/lxJ0SfX5pJ",
            "indices": [
              83,
              106
            ]
          }
        ]
      },
      "url": {
        "urls": [
          {
            "display_url": "github.com/vllm-project/v…",
            "expanded_url": "https://github.com/vllm-project/vllm",
            "url": "https://t.co/KmyOI0Gnbj",
            "indices": [
              0,
              23
            ]
          }
        ]
      }
    },
    "fastFollowersCount": 0,
    "favouritesCount": 503,
    "hasCustomTimelines": false,
    "isTranslator": false,
    "mediaCount": 177,
    "statusesCount": 736,
    "withheldInCountries": [],
    "affiliatesHighlightedLabel": {},
    "possiblySensitive": false,
    "pinnedTweetIds": [],
    "profile_bio": {
      "description": "A high-throughput and memory-efficient inference and serving engine for LLMs. Join https://t.co/lxJ0SfX5pJ to discuss together with the community!"
    },
    "isAutomated": false,
    "automatedBy": null
  },
  "extendedEntities": {
    "media": [
      {
        "display_url": "pic.x.com/WAEzc16Fbm",
        "expanded_url": "https://x.com/vllm_project/status/2002938138549682366/photo/1",
        "id_str": "2002938036129312768",
        "indices": [
          281,
          304
        ],
        "media_key": "3_2002938036129312768",
        "media_url_https": "https://pbs.twimg.com/media/G8vdiCabMAAWkci.jpg",
        "type": "photo",
        "url": "https://t.co/WAEzc16Fbm",
        "ext_media_availability": {
          "status": "Available"
        },
        "features": {
          "large": {
            "faces": [
              {
                "x": 239,
                "y": 387,
                "h": 66,
                "w": 66
              }
            ]
          },
          "medium": {
            "faces": [
              {
                "x": 219,
                "y": 355,
                "h": 60,
                "w": 60
              }
            ]
          },
          "small": {
            "faces": [
              {
                "x": 124,
                "y": 201,
                "h": 34,
                "w": 34
              }
            ]
          },
          "orig": {
            "faces": [
              {
                "x": 239,
                "y": 387,
                "h": 66,
                "w": 66
              }
            ]
          }
        },
        "sizes": {
          "large": {
            "h": 628,
            "w": 1306,
            "resize": "fit"
          },
          "medium": {
            "h": 577,
            "w": 1200,
            "resize": "fit"
          },
          "small": {
            "h": 327,
            "w": 680,
            "resize": "fit"
          },
          "thumb": {
            "h": 150,
            "w": 150,
            "resize": "crop"
          }
        },
        "original_info": {
          "height": 628,
          "width": 1306,
          "focus_rects": [
            {
              "x": 93,
              "y": 0,
              "w": 1121,
              "h": 628
            },
            {
              "x": 339,
              "y": 0,
              "w": 628,
              "h": 628
            },
            {
              "x": 378,
              "y": 0,
              "w": 551,
              "h": 628
            },
            {
              "x": 496,
              "y": 0,
              "w": 314,
              "h": 628
            },
            {
              "x": 0,
              "y": 0,
              "w": 1306,
              "h": 628
            }
          ]
        },
        "media_results": {
          "result": {
            "media_key": "3_2002938036129312768"
          }
        }
      }
    ]
  },
  "card": null,
  "place": {},
  "entities": {
    "hashtags": [],
    "symbols": [],
    "urls": [
      {
        "display_url": "docs.vllm.ai/projects/recip…",
        "expanded_url": "https://docs.vllm.ai/projects/recipes/en/latest/MiMo/MiMo-V2-Flash.html",
        "url": "https://t.co/MpUFWCPGjm",
        "indices": [
          737,
          760
        ]
      }
    ],
    "user_mentions": [
      {
        "id_str": "1914997266890579968",
        "name": "Xiaomi MiMo",
        "screen_name": "XiaomiMiMo",
        "indices": [
          771,
          782
        ]
      }
    ]
  },
  "quoted_tweet": null,
  "retweeted_tweet": null,
  "article": null
}