Datasets:

AgenticFilmLab
/

beyond-open-weights-fal-model-access-record

Languages:

English

Tags:

ai-filmmaking

synthetic-video

provenance

License:

Dataset card Files Files and versions

xet

Community

Dataset Viewer

Duplicate

Split (1)

train

The dataset viewer is not available for this split.

Cannot load the dataset split (in streaming mode) to extract the first rows.

Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
cell_id: string
scope: string
model_route: string
relative_path: string
sha256: string
bytes: int64
endpoint_id: string
request_file: string
video_width: int64
video_height: int64
fps: string
native_video_frames: string
duration_seconds: double
native_audio_streams: int64
native_audio_sample_rates: list<item: string>
  child 0, item: string
quoted_cost_usd: double
cost_basis: string
recorded_request_seconds: double
request_clock_scope: string
credentials_included: bool
voice_reference_included: bool
current_round: int64
files: list<item: struct<path: string, bytes: int64, sha256: string>>
  child 0, item: struct<path: string, bytes: int64, sha256: string>
      child 0, path: string
      child 1, bytes: int64
      child 2, sha256: string
main_cut: int64
historical_h3: int64
archived_wrong_brief: int64
raw_clip_count: int64
canary: int64
to
{'files': List({'path': Value('string'), 'bytes': Value('int64'), 'sha256': Value('string')}), 'raw_clip_count': Value('int64'), 'main_cut': Value('int64'), 'current_round': Value('int64'), 'archived_wrong_brief': Value('int64'), 'historical_h3': Value('int64'), 'canary': Value('int64'), 'voice_reference_included': Value('bool'), 'credentials_included': Value('bool')}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 147, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                File "/src/services/worker/src/worker/utils.py", line 127, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 483, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2951, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2461, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2486, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 547, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 430, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              cell_id: string
              scope: string
              model_route: string
              relative_path: string
              sha256: string
              bytes: int64
              endpoint_id: string
              request_file: string
              video_width: int64
              video_height: int64
              fps: string
              native_video_frames: string
              duration_seconds: double
              native_audio_streams: int64
              native_audio_sample_rates: list<item: string>
                child 0, item: string
              quoted_cost_usd: double
              cost_basis: string
              recorded_request_seconds: double
              request_clock_scope: string
              credentials_included: bool
              voice_reference_included: bool
              current_round: int64
              files: list<item: struct<path: string, bytes: int64, sha256: string>>
                child 0, item: struct<path: string, bytes: int64, sha256: string>
                    child 0, path: string
                    child 1, bytes: int64
                    child 2, sha256: string
              main_cut: int64
              historical_h3: int64
              archived_wrong_brief: int64
              raw_clip_count: int64
              canary: int64
              to
              {'files': List({'path': Value('string'), 'bytes': Value('int64'), 'sha256': Value('string')}), 'raw_clip_count': Value('int64'), 'main_cut': Value('int64'), 'current_round': Value('int64'), 'archived_wrong_brief': Value('int64'), 'historical_h3': Value('int64'), 'canary': Value('int64'), 'voice_reference_included': Value('bool'), 'credentials_included': Value('bool')}
              because column names don't match

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

Beyond Open Weights: Exploring Video Models on fal.ai

Agentic Film Lab's public experiment record for the v012, 15:09 episode. Six model families on fal; four tailored H3 requests are also run through RunPod Serverless. These are samples under documented conditions, not a universal ranking.

Complete scope

71 raw clips are preserved: 44 complete primaries in the final cut, 10 wrong-brief pottery attempts archived, 16 earlier H3 baseline tests, and one repair canary. The video opens with rally/deep-sea footage, moves through postal machinery, then ends its tests with corrected pottery. The ten original pottery requests asked for a small pouring lip rather than the intended continuous circular rim. They remain here rather than disappearing from the record. The repeated three-second plan at 2:51 in v011 was removed in v012 without removing narration, clips or full prompts.

Costs and clocks

The on-screen totals cover the current 54-take set, including its ten archived pottery attempts: 50 fal quotes/token estimates total $92.55, and four RunPod execution estimates total $3.96. Seedance uses a conservative token estimate. These are not settled invoices. The 16 earlier H3 takes and canary are a separate historical annex, not silently included in the 54-take table. Historical budgeting and partial provider billing remain separately labeled. Startup, idle, reference images and retained storage are not assumed free. Do not add overlapping execution estimates and endpoint bills.

Download and inspect

Complete evidence ZIP — all clips, request inputs, prompts, node graphs, two historical reference PNGs, scoped cost/timing records, captions and edit map.
Archive SHA-256
Clip index / full JSON index
Primary model/tool/provider sources
Reproduction and limitations / rights and exclusions

The raw clips retain native resolution, frame rate and audio. The final YouTube master is a 4K upscale of a 1080p composition; it is not native 4K generation. No model weights, creator voice reference or account credentials are included. No claim that all hosted model versions are closed-weight follows from their presence on fal.

Source clips

Clip	Scope	Output	Input
CANARY-REPAIR-S03	historical-canary	video	request
S01-T1-F	historical-H3-baseline	video	request
S01-T1-R	historical-H3-baseline	video	request
S01-T2-F	historical-H3-baseline	video	request
S01-T2-R	historical-H3-baseline	video	request
S02-T1-F	historical-H3-baseline	video	request
S02-T1-R	historical-H3-baseline	video	request
S02-T2-F	historical-H3-baseline	video	request
S02-T2-R	historical-H3-baseline	video	request
S03-RIM-R1-F	main-cut	video	request
S03-RIM-R1-G15	main-cut	video	request
S03-RIM-R1-K3TS	main-cut	video	request
S03-RIM-R1-OM11	main-cut	video	request
S03-RIM-R1-S	main-cut	video	request
S03-RIM-R1-X	main-cut	video	request
S03-T1-F	historical-H3-baseline	video	request
S03-T1-G15	archived-wrong-brief	video	request
S03-T1-K3TS	archived-wrong-brief	video	request
S03-T1-OM11	archived-wrong-brief	video	request
S03-T1-R	historical-H3-baseline	video	request
S03-T1-S	archived-wrong-brief	video	request
S03-T1-X	archived-wrong-brief	video	request
S03-T2-F	historical-H3-baseline	video	request
S03-T2-G15	archived-wrong-brief	video	request
S03-T2-K3TS	archived-wrong-brief	video	request
S03-T2-OM11	archived-wrong-brief	video	request
S03-T2-R	historical-H3-baseline	video	request
S03-T2-S	archived-wrong-brief	video	request
S03-T2-X	archived-wrong-brief	video	request
S03-TAIL-R1-F	main-cut	video	request
S03-TAIL-R1-G15	main-cut	video	request
S03-TAIL-R1-K3TS	main-cut	video	request
S03-TAIL-R1-OM11	main-cut	video	request
S03-TAIL-R1-S	main-cut	video	request
S03-TAIL-R1-X	main-cut	video	request
S03-TAIL-RP1	main-cut	video	request
S04-T1-F	historical-H3-baseline	video	request
S04-T1-G15	main-cut	video	request
S04-T1-K3TS	main-cut	video	request
S04-T1-OM11	main-cut	video	request
S04-T1-R	historical-H3-baseline	video	request
S04-T1-S	main-cut	video	request
S04-T1-X	main-cut	video	request
S04-T2-F	historical-H3-baseline	video	request
S04-T2-G15	main-cut	video	request
S04-T2-K3TS	main-cut	video	request
S04-T2-OM11	main-cut	video	request
S04-T2-R	historical-H3-baseline	video	request
S04-T2-S	main-cut	video	request
S04-T2-X	main-cut	video	request
S04-TAIL-R1-F	main-cut	video	request
S04-TAIL-R1-G15	main-cut	video	request
S04-TAIL-R1-K3TS	main-cut	video	request
S04-TAIL-R1-OM11	main-cut	video	request
S04-TAIL-R1-S	main-cut	video	request
S04-TAIL-R1-X	main-cut	video	request
S04-TAIL-RP1	main-cut	video	request
S05-TAIL-R1-F	main-cut	video	request
S05-TAIL-R1-G15	main-cut	video	request
S05-TAIL-R1-K3TS	main-cut	video	request
S05-TAIL-R1-OM11	main-cut	video	request
S05-TAIL-R1-S	main-cut	video	request
S05-TAIL-R1-X	main-cut	video	request
S05-TAIL-RP1	main-cut	video	request
S06-TAIL-R1-F	main-cut	video	request
S06-TAIL-R1-G15	main-cut	video	request
S06-TAIL-R1-K3TS	main-cut	video	request
S06-TAIL-R1-OM11	main-cut	video	request
S06-TAIL-R1-S	main-cut	video	request
S06-TAIL-R1-X	main-cut	video	request
S06-TAIL-RP1	main-cut	video	request

Downloads last month: 90

Total file size:

759 MB