← NeuralDrift Lab/BENCHMARK RESULT

qwen2.5.1-coder-7b-instruct

PASSNEURALDRIFT BENCHMARKED

Workload ND-AGENT-001 evaluated on NVIDIA GeForce RTX 5080 via lm_studio (0.3.9) using CUDA backend.

Compare this run →Download raw evidence JSON

Measured Telemetry

End-to-end agentic metrics capturing tool selection, argument execution, and task completion.

TASK COMPLETION TIME

0.66 s

End-to-end task time including agent reasoning & tools.

TOOL CALLS

1 / 1

Successful calls vs total attempted invocations.

TOOL LATENCY

0.01 ms

Average local tool execution resolution time.

TASK OUTCOME

SUCCESS

Validated objective result.

PEAK VRAM OBSERVED

10.57 GB

Dedicated GPU memory usage during agent run.

RETRIES REQUIRED

0

Error recovery turns needed.

Objective Validation

VALIDATOR: TOOL_CALL

VALIDATION PASSED

Successfully verified tool call "compute_memory_bandwidth".

OUTPUT RECEIVED

{"name":"compute_memory_bandwidth","arguments":{"busWidthBits":256,"memorySpeedGbps":28},"result":{"bandwidthGbps":896,"busWidthBits":256,"memorySpeedGbps":28},"latencyMs":0.01,"success":true}

Environment & Reproducibility

HARDWARE PROFILE

GPU	NVIDIA GeForce RTX 5080
Architecture	Blackwell
VRAM Capacity	15.92 GB
System RAM	61.64 GB
CPU	AMD Ryzen 9 9950X3D 16-Core Processor
Form Factor	DESKTOP

SOFTWARE & RUNTIME

Runtime	lm_studio (0.3.9)
Backend	CUDA 13.4
GPU Driver	616.92
Operating System	win32 10.0.26200
Model Quantization	GGUF / Q4_K_M
Context Limit	32768 tokens

CONFIGURATION FINGERPRINT

Deterministic SHA-256 hash across model, quantization, runtime, runtime version, GPU, driver, workload version, and generation settings. Two runs must share material variables to be directly comparable.

6b3f1ce29db8cc1abf3b4a381de01901eea5b6024f5e1e430dc89f24205a5629

Run ID: run-2026-09-17T13-28-06-342Z-nd-agent-001-qwen2_5_1-coder-7b-instruct-rtx-5080-61910625 · Recorded at: 2026-09-17T13:28:07.003Z · Benchmark Suite v1.0.0