Run code

Run a local file

No sandbox to create or manage. The language is an id or a code — both work.

boltz run --language python main.py
boltz run --language 100 main.py

Run inline code

boltz run --language python -c "print(1 + 1)"

Judge a solution

Test input, the expected answer, and the problem's limits. The exit status is the verdict: 0 when it is Accepted, 1 for a wrong answer, a time limit or a compile error, and the program's own code when it exits non-zero.

boltz run --language python solution.py \
  --stdin input.txt --expected output.txt \
  --time-limit 1s --memory-limit 64

boltz run --language cpp solution.cpp --stdin - < input.txt   # stdin from a pipe

Time and memory

boltz run --language python main.py --stats
# (Accepted · 0.012s CPU · 0.015s wall · 9216 KB)

The full response as JSON

--json prints the whole result — status, time, memory, stdout, stderr, compile output — exactly as the HTTP API returns it.

boltz run --language python main.py --json

Run a batch

Batch submission is available to paid users. Put 1–20 submissions in a JSON file. Each entry has its own source, language id, input, expected output and limits, in the same format as the HTTP API. The command waits for completion and prints JSON results in input order, retaining any per-item validation errors. It exits non-zero if any submission fails or is unavailable.

{
  "submissions": [
    {"source_code": "print(input())", "language_id": 100, "stdin": "hello", "expected_output": "hello"},
    {"source_code": "print(input())", "language_id": 100, "stdin": "world", "expected_output": "world"}
  ]
}

boltz run --batch jobs.json
boltz run --batch - < jobs.json
boltz run --batch jobs.json --stats --wait-timeout 5m

# Return tokens immediately; retrieve results later.
boltz run --batch jobs.json --wait=false
boltz run --tokens token1,token2
boltz run --tokens token1,token2 --wait=false   # current results, without waiting

Batch output is always JSON. An asynchronous submission returns the API's token/error array; waiting and token retrieval return an object with a submissions array. Waiting times out after 21 minutes by default and prints the last results so you can retrieve their tokens again. Use --base64-encoded for a file containing Base64 encoded text fields and to receive encoded results.

Available languages

boltz languages

Ids, codes and versions are listed on Supported languages.