dev.cocore.compute.provider

cocore.dev

Samples

100 randomly sampled records from the AT Protocol firehose

dev.cocore.compute.provider (100 samples) 
{
  "os": "macOS 26.5.1",
  "chip": "Apple M5",
  "$type": "dev.cocore.compute.provider",
  "ramGB": 24,
  "eCores": 6,
  "pCores": 4,
  "serving": true,
  "cpuCores": 10,
  "gpuCores": 10,
  "createdAt": "2026-06-19T22:40:26.369746Z",
  "priceList": [
    {
      "modelId": "mlx-community/Qwen2.5-0.5B-Instruct-4bit",
      "currency": "CC",
      "inputPricePerMTok": 1000000,
      "outputPricePerMTok": 1000000
    },
    {
      "modelId": "mlx-community/Qwen2.5-3B-Instruct-4bit",
      "currency": "CC",
      "inputPricePerMTok": 1000000,
      "outputPricePerMTok": 1000000
    },
    {
      "modelId": "stub",
      "currency": "CC",
      "inputPricePerMTok": 1000000,
      "outputPricePerMTok": 1000000
    }
  ],
  "trustLevel": "self-attested",
  "engineFault": {
    "at": "2026-06-19T22:40:26.341435Z",
    "code": "model-load-failed",
    "models": [
      "mlx-community/gemma-3-4b-it-qat-4bit"
    ],
    "message": "The inference engine for [mlx-community/gemma-3-4b-it-qat-4bit] did not come online after 3 attempts. The machine is online but only serving the no-op `stub` engine, so it won't be matched to real inference jobs. Most common cause: the model isn't in MLX format — cocore runs MLX weights (anything under `mlx-community/...`, or any repo carrying MLX 4-bit weights). A stock PyTorch / safetensors repo (e.g. `meta-llama/...`) will NOT load; look for an MLX conversion of it instead. Other causes: the model id is wrong or not on Hugging Face, the weight download failed, or the machine ran out of memory loading it. Fix: use an MLX-format model id, ensure the machine can reach Hugging Face, then start serving again. If it keeps failing, DM @cocore.dev on Bluesky with your machine label and we'll help."
  },
  "machineLabel": "aurora",
  "provisioning": false,
  "binaryVersion": "0.9.17",
  "modelIdentifier": "Mac17,3",
  "supportedModels": [
    "mlx-community/Qwen2.5-0.5B-Instruct-4bit",
    "mlx-community/Qwen2.5-3B-Instruct-4bit",
    "stub"
  ],
  "encryptionPubKey": "Pg/st8pgeThOhJH8pFTKi2NaIHpxmTKHtqw1pC5RYlA=",
  "attestationPubKey": "FsZK2EU/dH5/X/6qWRPSMM0L9+GNY1x6f4iRsmkngnslVRCatsNNZMPfjQeAUflTfF0P8KbMRHFRX67rojmnQQ=="
}

did:plc:jqtmhit665gpwjqpzclnq4mr | at://did:plc:jqtmhit665gpwjqpzclnq4mr/dev.cocore.compute.provider/3moo4ff6apl2k

Lexicon Garden

@