BLXBenchBLXBench UI
blxbench

Benchmark

Levels

Misc

DocsDownload blxbenchOur TestsPassSponsor / Partnership
DocsDownload blxbenchOur TestsPassSponsor / Partnership
BLXBenchBLXBench UI
blxbench

Benchmark

Levels

Misc

DocsDownload blxbenchOur TestsPassSponsor / Partnership
DocsDownload blxbenchOur TestsPassSponsor / Partnership
  1. Home
  2. Our Tests
  3. Multi Step
blxbench

Test fixture

Multi Step

Reasoninghardscorer: exact_match

Arithmetic, symbolic steps, and structured problem solving.

How it is scored

The model receives the prompt (and optional system message). The run uses scorer exact_match with the JSON configuration below. Pass/fail and partial credit are determined entirely by that scorer against the model output; no human grading.

User prompt
Compute ((32 + 17 + 10) * 2 - 8). Return only the number.
Scorer config
{
  "expected": "110"
}
Run parameters

temperature

0

max_tokens

32

timeout (s)

120

type

scored

file

reasoning_hard_03.json

← PreviousMulti Step
|
Next →Multi Step

BLXBench

Community driven leaderboardPublic benchmark runner — run in your environment, share results with the community.

© 2026 BLXBench by bitslix.com

ProvenanceAggregated from user runs
Scope1 / 7 / 372
Latestrun_1aa6fc / 7 / $0.00
TermsPrivacy