Self-reportedpublished

Cascade Agent 14B tool-use smoke test

No machine evidence attached yet.

Cascade Agent 14B @chen Jul 20, 2026

Manifest JSON Reproduce this Run

Speed

36.8 tok/s

Time to first token

444 ms

Peak memory

22.7 GB

Tool schema adherence

Useful

0 reproductions

Reproduction command

Use the exact revision and inspect the manifest before execution.

vllm serve agent-foundry/cascade-agent-14b --max-model-len 32768

Environment and configuration

The comparison context that makes the result interpretable.

Hardware

MachineRTX 4090 · 24 GB

CPUAMD Ryzen 9 7950X

AcceleratorNVIDIA GeForce RTX 4090

VRAM24 GB

System memory64 GB

Operating systemUbuntu 26.04

Architecturex64

Compute APICUDA 13.0

Configuration

RuntimevLLM 0.11.0

Model revision1b3f901

QuantizationQ4_K_M

Context length32768

Temperature0.2

Seed31

Batch size1

Portable Run manifest

Version 1.0 of the Apache-2.0 licensed Modelverse Run schema.

{
  "schemaVersion": "1.0",
  "id": "run_cascade_4090",
  "createdAt": "2026-07-20T13:00:00.000Z",
  "model": {
    "provider": "github",
    "repository": "agent-foundry/cascade-agent-14b",
    "revision": "1b3f901",
    "quantization": "Q4_K_M",
    "licenseSnapshot": "Custom model license"
  },
  "runtime": {
    "name": "vLLM",
    "version": "0.11.0",
    "command": "vllm serve agent-foundry/cascade-agent-14b --max-model-len 32768"
  },
  "hardware": {
    "os": "Ubuntu 26.04",
    "architecture": "x64",
    "cpu": "AMD Ryzen 9 7950X",
    "systemMemoryMb": 65536,
    "accelerators": [
      {
        "vendor": "NVIDIA",
        "model": "NVIDIA GeForce RTX 4090",
        "count": 1,
        "memoryMb": 24576,
        "driverVersion": "590.44",
        "computeApi": "cuda"
      }
    ]
  },
  "task": {
    "type": "tool-use"
  },
  "configuration": {
    "contextLength": 32768,
    "maxGeneratedTokens": 2048,
    "temperature": 0.2,
    "seed": 31,
    "batchSize": 1,
    "quantization": "Q4_K_M"
  },
  "metrics": {
    "timeToFirstTokenMs": 444,
    "tokensPerSecond": 36.8,
    "peakMemoryMb": 23244.8,
    "peakVramMb": 19353.6,
    "scores": {
      "Tool schema adherence": 72
    }
  },
  "evidence": {
    "machineCaptured": false,
    "redactionsApplied": [],
    "artifactUrls": []
  }
}

Notes and interpretation

Self-reported preview. License and performance require independent review.

Environment and configuration

The comparison context that makes the result interpretable.

Hardware

MachineRTX 4090 · 24 GB

CPUAMD Ryzen 9 7950X

AcceleratorNVIDIA GeForce RTX 4090

VRAM24 GB

System memory64 GB

Operating systemUbuntu 26.04

Architecturex64

Compute APICUDA 13.0

Configuration

RuntimevLLM 0.11.0

Model revision1b3f901

QuantizationQ4_K_M

Context length32768

Temperature0.2

Seed31

Batch size1

{ "schemaVersion": "1.0", "id": "run_cascade_4090", "createdAt": "2026-07-20T13:00:00.000Z", "model": { "provider": "github", "repository": "agent-foundry/cascade-agent-14b", "revision": "1b3f901", "quantization": "Q4_K_M", "licenseSnapshot": "Custom model license" }, "runtime": { "name": "vLLM", "version": "0.11.0", "command": "vllm serve agent-foundry/cascade-agent-14b --max-model-len 32768" }, "hardware": { "os": "Ubuntu 26.04", "architecture": "x64", "cpu": "AMD Ryzen 9 7950X", "systemMemoryMb": 65536, "accelerators": [ { "vendor": "NVIDIA", "model": "NVIDIA GeForce RTX 4090", "count": 1, "memoryMb": 24576, "driverVersion": "590.44", "computeApi": "cuda" } ] }, "task": { "type": "tool-use" }, "configuration": { "contextLength": 32768, "maxGeneratedTokens": 2048, "temperature": 0.2, "seed": 31, "batchSize": 1, "quantization": "Q4_K_M" }, "metrics": { "timeToFirstTokenMs": 444, "tokensPerSecond": 36.8, "peakMemoryMb": 23244.8, "peakVramMb": 19353.6, "scores": { "Tool schema adherence": 72 } }, "evidence": { "machineCaptured": false, "redactionsApplied": [], "artifactUrls": [] } }