openai/gpt-oss
File Size
The distribution of size of files (measured in lines of code).
File Size Overall
  • There are 111 files with 17,322 lines of code.
    • 3 very long files (3,697 lines of code)
    • 4 long files (2,196 lines of code)
    • 17 medium size files (5,779 lines of code)
    • 17 small files (2,617 lines of code)
    • 70 very small files (3,033 lines of code)
21% | 12% | 33% | 15% | 17%
Legend:
1001+
501-1000
201-500
101-200
1-100


explore: grouped by folders | grouped by size | sunburst | 3D view
File Size per Extension
1001+
501-1000
201-500
101-200
1-100
c63% | 13% | 13% | 3% | 6%
py16% | 7% | 41% | 11% | 22%
metal0% | 28% | 33% | 16% | 21%
cc0% | 54% | 0% | 21% | 23%
h0% | 0% | 33% | 46% | 19%
m0% | 0% | 100% | 0% | 0%
ts0% | 0% | 38% | 60% | 1%
hpp0% | 0% | 75% | 0% | 24%
jsonl0% | 0% | 0% | 0% | 100%
yaml0% | 0% | 0% | 0% | 100%
in0% | 0% | 0% | 0% | 100%
File Size per Logical Decomposition
primary
1001+
501-1000
201-500
101-200
1-100
gpt_oss22% | 13% | 33% | 13% | 16%
compatibility-test0% | 0% | 36% | 57% | 5%
gpt-oss-mcp-server0% | 0% | 0% | 0% | 100%
_build0% | 0% | 0% | 0% | 100%
.github0% | 0% | 0% | 0% | 100%
ROOT0% | 0% | 0% | 0% | 100%
Longest Files (Top 50)
File# lines# units
metal-kernels.c
in gpt_oss/metal/source
 1454 27
api_server.py
in gpt_oss/responses_api
 1233 11
context.c
in gpt_oss/metal/source
 1010 12
metal
moematmul.metal
in gpt_oss/metal/source
 627 -
simple_browser_tool.py
in gpt_oss/tools/simple_browser
 553 29
model.c
in gpt_oss/metal/source
 514 9
end-to-end-threadgroup.cc
in gpt_oss/metal/benchmark
 502 14
metal
matmul.metal
in gpt_oss/metal/source
 491 -
healthbench_eval.py
in gpt_oss/evals
 463 15
metal-kernels.h
in gpt_oss/metal/source/include/internal
 459 -
model.py
in gpt_oss/triton
 453 21
metal.m
in gpt_oss/metal/source
 423 19
model.py
in gpt_oss/torch
 410 20
apply_patch.py
in gpt_oss/tools
 402 24
chat.py
in gpt_oss
 326 2
docker_tool.py
in gpt_oss/tools/python_docker
 293 15
create-local-model.py
in gpt_oss/metal/scripts
 288 9
generate.c
in gpt_oss/metal/source
 285 7
metal.hpp
in gpt_oss/metal/source/include/internal
 284 35
runCase.ts
in compatibility-test
 284 6
metal
sdpa.metal
in gpt_oss/metal/source
 264 -
context.c
in gpt_oss/metal/python
 236 2
backend.py
in gpt_oss/tools/simple_browser
 210 4
page_contents.py
in gpt_oss/tools/simple_browser
 208 19
end-to-end.cc
in gpt_oss/metal/benchmark
 198 2
__main__.py
in gpt_oss/evals
 190 1
metal
sample.metal
in gpt_oss/metal/source
 188 -
metal
topk.metal
in gpt_oss/metal/source
 182 -
kernel-args.h
in gpt_oss/metal/source/include/internal
 179 -
index.ts
in compatibility-test
 177 5
attention.py
in gpt_oss/triton
 175 4
types.py
in gpt_oss/responses_api
 164 -
events.py
in gpt_oss/responses_api
 157 -
tokenizer.c
in gpt_oss/metal/python
 154 1
tools.ts
in compatibility-test
 153 2
model.h
in gpt_oss/metal/source/include/internal
 140 -
ollama.py
in gpt_oss/responses_api/inference
 128 5
analysis.ts
in compatibility-test
 116 2
metal.h
in gpt_oss/metal/source/include/internal
 109 -
functions.h
in gpt_oss/metal/include/gpt-oss
 104 -
uuid.h
in gpt_oss/metal/source/include/internal
 103 4
weights.py
in gpt_oss/torch
 99 5
gpqa_eval.py
in gpt_oss/evals
 92 3
stub.py
in gpt_oss/responses_api/inference
 92 2
macros.h
in gpt_oss/metal/source/include/internal
 90 -
browser_server.py
in gpt-oss-mcp-server
 90 2
report.py
in gpt_oss/evals
 89 5
tokenizer.c
in gpt_oss/metal/source
 89 7
build-system-prompt.py
in gpt-oss-mcp-server
 88 2
utils.py
in gpt_oss/responses_api
 87 1
Large Files That Change Often (Top 16)
To change a file, an AI coding agent (or a person) has to read it: these files are read the most for the changes in the past year (lines × changes).
File# lines≈ tokens
to read		# changes
(1y)		≈ lines edited
per change		lines read per
line changed
metal-kernels.c
in gpt_oss/metal/source
 1454 ~10k–30k 2 ~100 ~10
context.c
in gpt_oss/metal/source
 1010 ~7k–15k 2 ~100 ~10
api_server.py
in gpt_oss/responses_api
 1233 ~7k–20k 1 ~100 ~15
metal-kernels.h
in gpt_oss/metal/source/include/internal
 459 ~3k–7k 2 ~30 ~15
chat.py
in gpt_oss
 326 ~2k–5k 2 ~5 ~70
metal
moematmul.metal
in gpt_oss/metal/source
 627 ~3k–10k 1 ~200 ~3
model.c
in gpt_oss/metal/source
 514 ~3k–10k 1 ~15 ~30
metal
matmul.metal
in gpt_oss/metal/source
 491 ~3k–7k 1 ~100 ~5
backend.py
in gpt_oss/tools/simple_browser
 210 ~1k–3k 2 ~50 ~5
kernel-args.h
in gpt_oss/metal/source/include/internal
 179 ~1k–3k 2 ~15 ~10
context.c
in gpt_oss/metal/python
 236 ~1.5k–5k 1 ~20 ~10
model.h
in gpt_oss/metal/source/include/internal
 140 ~700–2k 1 ~7 ~20
browser_server.py
in gpt-oss-mcp-server
 90 ~500–1.5k 1 ~50 ~2
reference-system-prompt.py
in gpt-oss-mcp-server
 43 under 1k 2 ~15 ~3
metal
rope.metal
in gpt_oss/metal/source
 48 under 1k 1 ~10 ~5
metal
expert_routing_metadata.metal
in gpt_oss/metal/source
 36 under 1k 1 ~50 ~1

Tokens are lines × 7 to 14 (analysis.fileReadsForChanges). Lines edited per change are the file's lines added and deleted divided by its commits over the whole history. Agents read about 2,000 lines at a time: when a 17,000-line file was split into modules, the tokens an agent read for the same change fell by 83% (The Economic Benefit of Refactoring, 2026).

Files With Most Units (Top 50)
File# lines# units
metal.hpp
in gpt_oss/metal/source/include/internal
 284 35
simple_browser_tool.py
in gpt_oss/tools/simple_browser
 553 29
metal-kernels.c
in gpt_oss/metal/source
 1454 27
apply_patch.py
in gpt_oss/tools
 402 24
model.py
in gpt_oss/triton
 453 21
model.py
in gpt_oss/torch
 410 20
page_contents.py
in gpt_oss/tools/simple_browser
 208 19
metal.m
in gpt_oss/metal/source
 423 19
healthbench_eval.py
in gpt_oss/evals
 463 15
docker_tool.py
in gpt_oss/tools/python_docker
 293 15
end-to-end-threadgroup.cc
in gpt_oss/metal/benchmark
 502 14
context.c
in gpt_oss/metal/source
 1010 12
api_server.py
in gpt_oss/responses_api
 1233 11
backend.py
in _build/gpt_oss_build_backend
 75 11
model.c
in gpt_oss/metal/source
 514 9
create-local-model.py
in gpt_oss/metal/scripts
 288 9
generate.c
in gpt_oss/metal/source
 285 7
datatype.hpp
in gpt_oss/metal/source/include/internal
 74 7
tokenizer.c
in gpt_oss/metal/source
 89 7
tool.py
in gpt_oss/tools
 52 6
math.h
in gpt_oss/metal/source/include/internal
 31 6
runCase.ts
in compatibility-test
 284 6
aime_eval.py
in gpt_oss/evals
 81 5
report.py
in gpt_oss/evals
 89 5
weights.py
in gpt_oss/torch
 99 5
ollama.py
in gpt_oss/responses_api/inference
 128 5
index.ts
in compatibility-test
 177 5
backend.py
in gpt_oss/tools/simple_browser
 210 4
uuid.h
in gpt_oss/metal/source/include/internal
 103 4
attention.py
in gpt_oss/triton
 175 4
gpqa_eval.py
in gpt_oss/evals
 92 3
responses_sampler.py
in gpt_oss/evals
 71 3
80 3
moe.py
in gpt_oss/triton
 49 3
triton.py
in gpt_oss/responses_api/inference
 75 3
vllm.py
in gpt_oss/responses_api/inference
 43 3
transformers.py
in gpt_oss/responses_api/inference
 27 3
abcd_grader.py
in gpt_oss/evals
 78 2
basic_eval.py
in gpt_oss/evals
 30 2
types.py
in gpt_oss/evals
 34 2
end-to-end.cc
in gpt_oss/metal/benchmark
 198 2
model.c
in gpt_oss/metal/python
 77 2
context.c
in gpt_oss/metal/python
 236 2
utils.py
in gpt_oss/torch
 28 2
token_generator.py
in gpt_oss/vllm
 44 2
chat.py
in gpt_oss
 326 2
stub.py
in gpt_oss/responses_api/inference
 92 2
build-system-prompt.py
in gpt-oss-mcp-server
 88 2
browser_server.py
in gpt-oss-mcp-server
 90 2
tools.ts
in compatibility-test
 153 2
Files With Long Lines (Top 27)

There are 27 files with lines longer than 120 characters. In total, there are 221 long lines.

File# lines# units# long lines
metal-kernels.c
in gpt_oss/metal/source
 1454 27 78
model.c
in gpt_oss/metal/source
 514 9 28
jsonl
cases.jsonl
in compatibility-test
 30 - 21
context.c
in gpt_oss/metal/source
 1010 12 19
end-to-end-threadgroup.cc
in gpt_oss/metal/benchmark
 502 14 14
uuid.h
in gpt_oss/metal/source/include/internal
 103 4 14
metal
sdpa.metal
in gpt_oss/metal/source
 264 - 9
healthbench_eval.py
in gpt_oss/evals
 463 15 5
metal
moematmul.metal
in gpt_oss/metal/source
 627 - 4
end-to-end.cc
in gpt_oss/metal/benchmark
 198 2 3
generate.c
in gpt_oss/metal/source
 285 7 3
create-local-model.py
in gpt_oss/metal/scripts
 288 9 3
chat.py
in gpt_oss
 326 2 3
metal
matmul.metal
in gpt_oss/metal/source
 491 - 2
moe.py
in gpt_oss/triton
 49 3 2
weights.py
in gpt_oss/torch
 99 5 2
simple_browser_tool.py
in gpt_oss/tools/simple_browser
 553 29 1
context.c
in gpt_oss/metal/python
 236 2 1
tokenizer.c
in gpt_oss/metal/python
 154 1 1
metal.m
in gpt_oss/metal/source
 423 19 1
metal
sample.metal
in gpt_oss/metal/source
 188 - 1
generate.py
in gpt_oss
 83 1 1
attention.py
in gpt_oss/triton
 175 4 1
model.py
in gpt_oss/triton
 453 21 1
token_generator.py
in gpt_oss/vllm
 44 2 1
runCase.ts
in compatibility-test
 284 6 1
tools.ts
in compatibility-test
 153 2 1
Correlations

File Size vs. Commits (all time): 63 points

gpt_oss/tools/simple_browser/backend.py x: 3 commits (all time) y: 210 lines of code gpt-oss-mcp-server/browser_server.py x: 3 commits (all time) y: 90 lines of code gpt-oss-mcp-server/reference-system-prompt.py x: 4 commits (all time) y: 43 lines of code gpt_oss/chat.py x: 5 commits (all time) y: 326 lines of code gpt_oss/responses_api/api_server.py x: 7 commits (all time) y: 1233 lines of code gpt_oss/metal/python/context.c x: 4 commits (all time) y: 236 lines of code gpt_oss/metal/source/context.c x: 13 commits (all time) y: 1010 lines of code gpt_oss/metal/source/include/internal/kernel-args.h x: 7 commits (all time) y: 179 lines of code gpt_oss/metal/source/include/internal/metal-kernels.h x: 7 commits (all time) y: 459 lines of code gpt_oss/metal/source/matmul.metal x: 4 commits (all time) y: 491 lines of code gpt_oss/metal/source/metal-kernels.c x: 8 commits (all time) y: 1454 lines of code gpt_oss/metal/source/rope.metal x: 3 commits (all time) y: 48 lines of code gpt_oss/metal/source/expert_routing_metadata.metal x: 1 commits (all time) y: 36 lines of code gpt_oss/metal/source/include/internal/model.h x: 12 commits (all time) y: 140 lines of code gpt_oss/metal/source/model.c x: 10 commits (all time) y: 514 lines of code gpt_oss/metal/source/moematmul.metal x: 3 commits (all time) y: 627 lines of code gpt_oss/responses_api/types.py x: 4 commits (all time) y: 164 lines of code gpt_oss/tools/python_docker/docker_tool.py x: 6 commits (all time) y: 293 lines of code gpt_oss/metal/benchmark/end-to-end-threadgroup.cc x: 2 commits (all time) y: 502 lines of code gpt_oss/metal/benchmark/end-to-end.cc x: 7 commits (all time) y: 198 lines of code gpt_oss/metal/include/gpt-oss/functions.h x: 5 commits (all time) y: 104 lines of code gpt_oss/metal/python/model.c x: 2 commits (all time) y: 77 lines of code gpt_oss/metal/source/generate.c x: 4 commits (all time) y: 285 lines of code gpt_oss/responses_api/events.py x: 2 commits (all time) y: 157 lines of code gpt_oss/metal/scripts/create-local-model.py x: 1 commits (all time) y: 288 lines of code gpt_oss/metal/source/gather_and_accumulate.metal x: 1 commits (all time) y: 66 lines of code gpt_oss/metal/source/sdpa.metal x: 3 commits (all time) y: 264 lines of code gpt_oss/metal/benchmark/f32-bf16w-rmsnorm.cc x: 1 commits (all time) y: 82 lines of code gpt_oss/metal/source/accumulate.metal x: 1 commits (all time) y: 53 lines of code gpt_oss/metal/source/embeddings.metal x: 1 commits (all time) y: 24 lines of code gpt_oss/metal/source/rmsnorm.metal x: 1 commits (all time) y: 50 lines of code gpt_oss/metal/source/sample.metal x: 2 commits (all time) y: 188 lines of code gpt_oss/metal/source/topk.metal x: 1 commits (all time) y: 182 lines of code gpt_oss/responses_api/inference/metal.py x: 2 commits (all time) y: 23 lines of code gpt_oss/evals/responses_sampler.py x: 2 commits (all time) y: 71 lines of code gpt_oss/tools/simple_browser/__init__.py x: 1 commits (all time) y: 7 lines of code gpt_oss/metal/source/metal.m x: 2 commits (all time) y: 423 lines of code gpt_oss/metal/source/include/internal/math.h x: 2 commits (all time) y: 31 lines of code gpt_oss/metal/source/include/internal/metal.h x: 1 commits (all time) y: 109 lines of code gpt_oss/metal/source/include/internal/metal.hpp x: 1 commits (all time) y: 284 lines of code gpt_oss/triton/attention.py x: 2 commits (all time) y: 175 lines of code gpt_oss/evals/__main__.py x: 3 commits (all time) y: 190 lines of code gpt_oss/responses_api/inference/ollama.py x: 2 commits (all time) y: 128 lines of code gpt-oss-mcp-server/build-system-prompt.py x: 2 commits (all time) y: 88 lines of code gpt_oss/generate.py x: 2 commits (all time) y: 83 lines of code gpt_oss/tools/simple_browser/page_contents.py x: 2 commits (all time) y: 208 lines of code gpt_oss/evals/healthbench_eval.py x: 3 commits (all time) y: 463 lines of code compatibility-test/analysis.ts x: 1 commits (all time) y: 116 lines of code compatibility-test/cases.jsonl x: 1 commits (all time) y: 30 lines of code compatibility-test/index.ts x: 1 commits (all time) y: 177 lines of code compatibility-test/tools.ts x: 1 commits (all time) y: 153 lines of code gpt_oss/tools/simple_browser/simple_browser_tool.py x: 1 commits (all time) y: 553 lines of code MANIFEST.in x: 1 commits (all time) y: 1 lines of code
1454.0
lines of code
  min: 1.0
  average: 222.68
  25th percentile: 50.0
  median: 140.0
  75th percentile: 284.0
  max: 1454.0
0 13.0
commits (all time)
min: 1.0 | average: 2.92 | 25th percentile: 1.0 | median: 2.0 | 75th percentile: 4.0 | max: 13.0

File Size vs. Contributors (all time): 63 points

gpt_oss/tools/simple_browser/backend.py x: 3 contributors (all time) y: 210 lines of code gpt-oss-mcp-server/browser_server.py x: 3 contributors (all time) y: 90 lines of code gpt-oss-mcp-server/reference-system-prompt.py x: 4 contributors (all time) y: 43 lines of code gpt_oss/chat.py x: 4 contributors (all time) y: 326 lines of code gpt_oss/responses_api/api_server.py x: 4 contributors (all time) y: 1233 lines of code gpt_oss/metal/python/context.c x: 2 contributors (all time) y: 236 lines of code gpt_oss/metal/source/context.c x: 2 contributors (all time) y: 1010 lines of code gpt_oss/metal/source/include/internal/kernel-args.h x: 2 contributors (all time) y: 179 lines of code gpt_oss/metal/source/include/internal/metal-kernels.h x: 2 contributors (all time) y: 459 lines of code gpt_oss/metal/source/matmul.metal x: 2 contributors (all time) y: 491 lines of code gpt_oss/metal/source/metal-kernels.c x: 2 contributors (all time) y: 1454 lines of code gpt_oss/metal/source/rope.metal x: 2 contributors (all time) y: 48 lines of code gpt_oss/metal/source/expert_routing_metadata.metal x: 1 contributors (all time) y: 36 lines of code gpt_oss/metal/source/include/internal/model.h x: 2 contributors (all time) y: 140 lines of code gpt_oss/metal/source/model.c x: 2 contributors (all time) y: 514 lines of code gpt_oss/metal/source/moematmul.metal x: 2 contributors (all time) y: 627 lines of code gpt_oss/responses_api/types.py x: 2 contributors (all time) y: 164 lines of code gpt_oss/tools/python_docker/docker_tool.py x: 4 contributors (all time) y: 293 lines of code gpt_oss/metal/benchmark/end-to-end-threadgroup.cc x: 1 contributors (all time) y: 502 lines of code gpt_oss/metal/benchmark/end-to-end.cc x: 2 contributors (all time) y: 198 lines of code gpt_oss/metal/include/gpt-oss/functions.h x: 4 contributors (all time) y: 104 lines of code gpt_oss/metal/python/model.c x: 2 contributors (all time) y: 77 lines of code gpt_oss/metal/source/generate.c x: 2 contributors (all time) y: 285 lines of code gpt_oss/responses_api/events.py x: 1 contributors (all time) y: 157 lines of code gpt_oss/metal/scripts/create-local-model.py x: 1 contributors (all time) y: 288 lines of code gpt_oss/metal/source/gather_and_accumulate.metal x: 1 contributors (all time) y: 66 lines of code gpt_oss/metal/source/sdpa.metal x: 1 contributors (all time) y: 264 lines of code gpt_oss/metal/benchmark/f32-bf16w-rmsnorm.cc x: 1 contributors (all time) y: 82 lines of code gpt_oss/metal/source/accumulate.metal x: 1 contributors (all time) y: 53 lines of code gpt_oss/metal/source/embeddings.metal x: 1 contributors (all time) y: 24 lines of code gpt_oss/metal/source/rmsnorm.metal x: 1 contributors (all time) y: 50 lines of code gpt_oss/metal/source/sample.metal x: 1 contributors (all time) y: 188 lines of code gpt_oss/metal/source/topk.metal x: 1 contributors (all time) y: 182 lines of code gpt_oss/evals/responses_sampler.py x: 2 contributors (all time) y: 71 lines of code gpt_oss/tools/simple_browser/__init__.py x: 1 contributors (all time) y: 7 lines of code gpt_oss/metal/source/metal.m x: 1 contributors (all time) y: 423 lines of code gpt_oss/metal/source/include/internal/math.h x: 1 contributors (all time) y: 31 lines of code gpt_oss/metal/source/include/internal/metal.h x: 1 contributors (all time) y: 109 lines of code gpt_oss/metal/source/include/internal/metal.hpp x: 1 contributors (all time) y: 284 lines of code gpt_oss/triton/attention.py x: 2 contributors (all time) y: 175 lines of code gpt_oss/evals/__main__.py x: 2 contributors (all time) y: 190 lines of code gpt_oss/responses_api/inference/ollama.py x: 2 contributors (all time) y: 128 lines of code gpt-oss-mcp-server/build-system-prompt.py x: 2 contributors (all time) y: 88 lines of code gpt_oss/generate.py x: 2 contributors (all time) y: 83 lines of code gpt_oss/tools/simple_browser/page_contents.py x: 2 contributors (all time) y: 208 lines of code gpt_oss/evals/healthbench_eval.py x: 3 contributors (all time) y: 463 lines of code compatibility-test/analysis.ts x: 1 contributors (all time) y: 116 lines of code compatibility-test/index.ts x: 1 contributors (all time) y: 177 lines of code compatibility-test/tools.ts x: 1 contributors (all time) y: 153 lines of code gpt-oss-mcp-server/python_server.py x: 2 contributors (all time) y: 24 lines of code gpt_oss/tools/simple_browser/simple_browser_tool.py x: 1 contributors (all time) y: 553 lines of code gpt_oss/evals/chat_completions_sampler.py x: 1 contributors (all time) y: 80 lines of code MANIFEST.in x: 1 contributors (all time) y: 1 lines of code
1454.0
lines of code
  min: 1.0
  average: 222.68
  25th percentile: 50.0
  median: 140.0
  75th percentile: 284.0
  max: 1454.0
0 4.0
contributors (all time)
min: 1.0 | average: 1.7 | 25th percentile: 1.0 | median: 1.0 | 75th percentile: 2.0 | max: 4.0

File Size vs. Commits (30 days): 0 points

No data for "commits (30d)" vs. "lines of code".

File Size vs. Contributors (30 days): 0 points

No data for "contributors (30d)" vs. "lines of code".

File Size vs. Commits (90 days): 1 points

gpt_oss/tools/simple_browser/backend.py x: 1 commits (90d) y: 210 lines of code
210.0
lines of code
  min: 210.0
  average: 210.0
  25th percentile: 210.0
  median: 210.0
  75th percentile: 210.0
  max: 210.0
0 1.0
commits (90d)
min: 1.0 | average: 1.0 | 25th percentile: 1.0 | median: 1.0 | 75th percentile: 1.0 | max: 1.0

File Size vs. Contributors (90 days): 1 points

gpt_oss/tools/simple_browser/backend.py x: 1 contributors (90d) y: 210 lines of code
210.0
lines of code
  min: 210.0
  average: 210.0
  25th percentile: 210.0
  median: 210.0
  75th percentile: 210.0
  max: 210.0
0 1.0
contributors (90d)
min: 1.0 | average: 1.0 | 25th percentile: 1.0 | median: 1.0 | 75th percentile: 1.0 | max: 1.0
About This Analysis
  • File size measurements show the distribution of size of files.
  • Files are classified in four categories based on their size (lines of code): 1-100 (very small files), 101-200 (small files), 201-500 (medium size files), 501-1000 (long files), 1001+(very long files).
  • It is a good practice to keep files small. Long files may become "bloaters", code that have increased to such gargantuan proportions that they are hard to work with.
Learn more...