Experimental run record

Run df7df63d

An inspectable execution record: configuration, deterministic outcomes, attempts, artefacts, and failure traces. Comparison applies across the configurations in this run only.

· pack geoworkbench-0.1 · accepted · sealed snapshot 0cbcce31

← All runs

Trials
12
Accepted
6
Configurations
4
Tasks
3
Cost USD
0.02673

In shortKimi K2.6 led this run: 2/3 tasks accepted (66.7%) for 0.0039 USD total. This ranking holds inside this run only — it is not a general model ranking.

Comparison across the configurations in this run only

JSON · CSV
Ranked by accepted rate, then total cost. Not comparable across runs or pack versions.
#ModelArchitectureAcceptedRateFirst-attempt successFalse-success rejectedCost USDCost / acceptedMedian msp95 ms
1Kimi K2.6single-shot2/366.7%66.7%10.00390.002028078 ms42037 ms
2GLM-4.7 Flashsingle-shot2/366.7%66.7%10.00460.002335526 ms71853 ms
3GLM-4.7 Flashsingle-retry1/333.3%33.3%20.00900.009075372 ms83968 ms
4Kimi K2.6single-retry1/333.3%33.3%20.00910.009173802 ms95205 ms

Trials

attempt traces expand beneath each trial
Trial outcomes are deterministic verifier results, not model-judge opinions.
TaskConfigurationRetryResultAttemptsCost USDLatency ms
crs-reproject-munichGLM-4.7 Flashsingle-shotaccepted10.0015 USD35526 ms
Attempt trace undefined
# attempt undefined — model: @cf/zai-org/glm-4.7-flash — exit 0 — accepted — cost 0.0015 USD — 35526 ms
# verifier: PASS — CRS reproject to WGS84 verified
# artifact: download (591 bytes)
import geopandas as gpd
import pyproj

# Read the input file
gdf = gpd.read_file("input.geojson")

# Diagnose the CRS
# The coordinates are large (millions), indicating a projected CRS in metres.
# Based on the task description, we determine this is EPSG:3857 (Web Mercator).
gdf.crs = "EPSG:3857"

# Reproject to WGS84 (EPSG:4326)
gdf_wgs84 = gdf.to_crs("EPSG:4326")

# Write the result to output.geojson
gdf_wgs84.to_file("output.geojson", driver="GeoJSON")
crs-reproject-munichGLM-4.7 Flashsingle-retryrejected20.0036 USD83968 ms
Attempt trace undefined
# attempt undefined — model: @cf/zai-org/glm-4.7-flash — exit 0 — rejected — cost 0.0019 USD — 40707 ms
# verifier: FAIL — verifier did not emit JSON; stdout tail: 
# no artifact produced

Attempt trace undefined
# attempt undefined — model: @cf/zai-org/glm-4.7-flash — exit 0 — rejected — cost 0.0017 USD — 83968 ms
# verifier: FAIL — output not in WGS84 Munich coords
# artifact: download (582 bytes)
import geopandas as gpd

# Load the input file
gdf = gpd.read_file("input.geojson")

# The coordinates are large (millions), indicating a projected CRS.
# As per the task instructions, we determine this is EPSG:3857.
if gdf.crs is None:
    gdf.crs = "EPSG:3857"

# Reproject to WGS84 (EPSG:4326)
gdf_wgs84 = gdf.to_crs("EPSG:4326")

# Write the output
gdf_wgs84.to_file("output.geojson", driver="GeoJSON")
raster-zonal-meanGLM-4.7 Flashsingle-shotrejected10.0019 USD71853 ms
Attempt trace undefined
# attempt undefined — model: @cf/zai-org/glm-4.7-flash — exit 0 — rejected — cost 0.0019 USD — 71853 ms
# verifier: FAIL — stats.json unreadable
# no artifact produced

raster-zonal-meanGLM-4.7 Flashsingle-retryrejected20.0039 USD75372 ms
Attempt trace undefined
# attempt undefined — model: @cf/zai-org/glm-4.7-flash — exit 1 — rejected — cost 0.0017 USD — 32013 ms
# verifier: FAIL — stats.json unreadable
# no artifact produced
import rasterio
import numpy as np
import json
from shapely import geometry

def main():
    # Load the raster
    with rasterio.open('grid.asc') as src:
        data = src.read(1)
        transform = src.transform
        shape = src.shape
        nodata = src.nodata

    # Load the polygon geometry
    with open('zone.geojson', 'r') as f:
        geojson_data = json.load(f)
    
    # Extract geometry from GeoJSON feature
    geom = geometry.shape(geojson_data['features'][0]['geometry'])

    # Create a boolean mask for pixels inside the polygon
    mask = rasterio.features.geometry_mask(
        [geom],
        transform=transform,
        out_shape=shape,
        invert=False
    )

    # Apply the mask to the raster data
    masked_values = data[mask]

    # Filter out nodata values
    if nodata is not None:
        masked_values = masked_values[masked_values != nodata]

    # Filter out NaN values
    masked_values = masked_values[~np.isnan(masked_values)]

    # Calculate the mean
    if len(masked_values) > 0:
        mean_val = np.mean(masked_values)
    else:
        mean_val = None

    # Write the result to stats.json
    result = {"mean": mean_val}
    with open('stats.json', 'w') as f:
        json.dump(result, f)

if __name__ == "__main__":
    main()

--- stderr ---
Traceback (most recent call last):
  File "/tmp/arena/solution.py", line 51, in <module>
    main()
  File "/tmp/arena/solution.py", line 22, in main
    mask = rasterio.features.geometry_mask(
           ^^^^^^^^^^^^^^^^^
AttributeError: module 'rasterio' has no attribute 'features'. Did you mean: '_features'?
Attempt trace undefined
# attempt undefined — model: @cf/zai-org/glm-4.7-flash — exit 0 — rejected — cost 0.0022 USD — 75372 ms
# verifier: FAIL — stats.json unreadable
# no artifact produced

vector-buffer-1kmGLM-4.7 Flashsingle-shotaccepted10.0013 USD24848 ms
Attempt trace undefined
# attempt undefined — model: @cf/zai-org/glm-4.7-flash — exit 0 — accepted — cost 0.0013 USD — 24848 ms
# verifier: PASS — 1000m buffers verified
# artifact: download (5935 bytes)
import geopandas as gpd
from pyproj import CRS

# Read input data
gdf = gpd.read_file("sites.geojson")

# Define coordinate reference systems
src_crs = CRS.from_epsg(4326)
tgt_crs = CRS.from_epsg(3857)

# Reproject to projected CRS for accurate buffering
gdf_proj = gdf.to_crs(tgt_crs)

# Create 1000m buffers
gdf_proj['geometry'] = gdf_proj.buffer(1000)

# Reproject back to WGS84
gdf_final = gdf_proj.to_crs(src_crs)

# Write output
gdf_final.to_file("buffers.geojson", driver="GeoJSON")
vector-buffer-1kmGLM-4.7 Flashsingle-retryaccepted10.0015 USD30656 ms
Attempt trace undefined
# attempt undefined — model: @cf/zai-org/glm-4.7-flash — exit 0 — accepted — cost 0.0015 USD — 30656 ms
# verifier: PASS — 1000m buffers verified
# artifact: download (5903 bytes)
import geopandas as gpd

# Read the input file
gdf = gpd.read_file('sites.geojson')

# Define coordinate reference systems
# Source: WGS84 (EPSG:4326)
# Target: Web Mercator (EPSG:3857) for accurate meter-based buffering
src_crs = 'EPSG:4326'
dst_crs = 'EPSG:3857'

# Reproject to projected CRS to ensure 1000m distance is accurate
gdf_projected = gdf.to_crs(dst_crs)

# Create 1000m buffers
gdf_projected['buffer'] = gdf_projected.geometry.buffer(1000)

# Extract buffer geometries and reproject back to WGS84
buffers_gdf = gpd.GeoDataFrame(geometry=gdf_projected['buffer'], crs=dst_crs)
buffers_gdf = buffers_gdf.to_crs(src_crs)

# Write the output file
buffers_gdf.to_file('buffers.geojson', driver='GeoJSON')
crs-reproject-munichKimi K2.6single-shotaccepted10.0009 USD20975 ms
Attempt trace undefined
# attempt undefined — model: @cf/moonshotai/kimi-k2.6 — exit 0 — accepted — cost 0.0009 USD — 20975 ms
# verifier: PASS — CRS reproject to WGS84 verified
# artifact: download (591 bytes)
import geopandas as gpd

gdf = gpd.read_file("input.geojson")
gdf = gdf.set_crs(epsg=3857, allow_override=True)
gdf = gdf.to_crs(epsg=4326)
gdf.to_file("output.geojson", driver="GeoJSON")
crs-reproject-munichKimi K2.6single-retryrejected20.0039 USD73802 ms
Attempt trace undefined
# attempt undefined — model: @cf/moonshotai/kimi-k2.6 — exit 0 — rejected — cost 0.0019 USD — 41038 ms
# verifier: FAIL — verifier did not emit JSON; stdout tail: 
# no artifact produced

Attempt trace undefined
# attempt undefined — model: @cf/moonshotai/kimi-k2.6 — exit 0 — rejected — cost 0.0020 USD — 73802 ms
# verifier: FAIL — verifier did not emit JSON; stdout tail: 
# no artifact produced

raster-zonal-meanKimi K2.6single-shotrejected10.0019 USD42037 ms
Attempt trace undefined
# attempt undefined — model: @cf/moonshotai/kimi-k2.6 — exit 0 — rejected — cost 0.0019 USD — 42037 ms
# verifier: FAIL — stats.json unreadable
# no artifact produced

raster-zonal-meanKimi K2.6single-retryrejected20.0038 USD95205 ms
Attempt trace undefined
# attempt undefined — model: @cf/moonshotai/kimi-k2.6 — exit 0 — rejected — cost 0.0019 USD — 45507 ms
# verifier: FAIL — stats.json unreadable
# no artifact produced

Attempt trace undefined
# attempt undefined — model: @cf/moonshotai/kimi-k2.6 — exit 0 — rejected — cost 0.0019 USD — 95205 ms
# verifier: FAIL — stats.json unreadable
# no artifact produced

vector-buffer-1kmKimi K2.6single-shotaccepted10.0012 USD28078 ms
Attempt trace undefined
# attempt undefined — model: @cf/moonshotai/kimi-k2.6 — exit 0 — accepted — cost 0.0012 USD — 28078 ms
# verifier: PASS — 1000m buffers verified
# artifact: download (5935 bytes)
import geopandas as gpd

gdf = gpd.read_file("sites.geojson")
gdf_proj = gdf.to_crs("EPSG:3857")
gdf_proj["geometry"] = gdf_proj.buffer(1000)
gdf_out = gdf_proj.to_crs("EPSG:4326")
gdf_out.to_file("buffers.geojson", driver="GeoJSON")
vector-buffer-1kmKimi K2.6single-retryaccepted10.0014 USD31172 ms
Attempt trace undefined
# attempt undefined — model: @cf/moonshotai/kimi-k2.6 — exit 0 — accepted — cost 0.0014 USD — 31172 ms
# verifier: PASS — 1000m buffers verified
# artifact: download (5935 bytes)
import geopandas as gpd

sites = gpd.read_file("sites.geojson")
if sites.crs is None:
    sites.set_crs(epsg=4326, inplace=True)

buffers = sites.to_crs(epsg=3857)
buffers.geometry = buffers.buffer(1000)
buffers = buffers.to_crs(epsg=4326)

buffers.to_file("buffers.geojson", driver="GeoJSON")

Cite this record

Axis Spatial. "Experimental run df7df63d." Axis Spatial Arena, run df7df63d-6f59-4b78-85fe-65676d5f7543. Benchmark pack geoworkbench-0.1. https://www.axisspatial.com/arena/runs/df7df63d-6f59-4b78-85fe-65676d5f7543 (accessed 19 Sep 2026).
@misc{axis-spatial-arena-run-df7df63d,
  author = {{Axis Spatial}},
  title = {Experimental run df7df63d},
  howpublished = {Axis Spatial Arena, run: https://www.axisspatial.com/arena/runs/df7df63d-6f59-4b78-85fe-65676d5f7543},
  note = {Benchmark pack geoworkbench-0.1},
  year = {2026}
}

Glossary

full protocol in the methodology
Pack version
The pinned set of task definitions a run executed. Results from different pack versions are never compared directly.
Configuration
A complete comparison unit: model, prompt version, tools, retry policy, and pinned task, data, and verifier versions. A model name alone is not comparable.
Lane
One configured route through which a model is reached. Duplicate upstream models across lanes count as adapter evidence, not extra capability.
False-success rejection
A trial whose output looked plausible but failed independent deterministic checks, such as an empty file or a fabricated value.
Token-cost proxy
Cost estimated from reported token usage and list prices at recording time; actual billing may differ.
Sealed snapshot
An immutable, content-addressed copy of a leaderboard computed from a finished run.