TinyBench · open benchmark · Aug 2026

Frontier LLMs are
trivially wrong.

We gave 8 frontier models 40 tiny, exact tasks — hash a string, cross a half-hour DST boundary, amortize a 289-month loan. Unaided they averaged 68%. With TinyFn's MCP tools the same models scored 99%, and 6 of 8 were perfect.

68%unaided accuracy
99%with TinyFn tools
0/8solved MD5, SHA-256 or HMAC unaided
640graded evaluations

Leaderboard

Same model, same prompt, same grader — the only difference is tool access.

claude-opus-5anthropic
+5 tasks with tools
unaided
35/40 · 88%
with TinyFn
40/40 · 100%
gpt-5.5openai
+6 tasks with tools
unaided
34/40 · 85%
with TinyFn
40/40 · 100%
grok-4.6x-ai
+6 tasks with tools
unaided
34/40 · 85%
with TinyFn
40/40 · 100%
gemini-3.1-pro-previewgoogle
+7 tasks with tools
unaided
33/40 · 82%
with TinyFn
40/40 · 100%
claude-sonnet-5anthropic
+7 tasks with tools
unaided
32/40 · 80%
with TinyFn
39/40 · 98%
gemini-3.7-flashgoogle
+10 tasks with tools
unaided
30/40 · 75%
with TinyFn
40/40 · 100%
claude-haiku-4.5anthropic
+27 tasks with tools
unaided
11/40 · 28%
with TinyFn
38/40 · 95%
gpt-5.4-miniopenai
+30 tasks with tools
unaided
10/40 · 25%
with TinyFn
40/40 · 100%

The receipts

Wrong answers aren't approximate — they're confidently, plausibly wrong.

gpt-5.5 · HMAC-SHA256 — fabricated a perfectly plausible digest
it saidc0b566d7ae8a9df97a3f8a9c04652c8eea4ea751c9c32b4143f9840c9bf69ebd
truth700a72e1c8e7eb1aed4cfc76f4d45d0a7a8553cb528bf4b715b4ba03af2da375
claude-opus-5 · MD5 — invented a hash rather than admit it can't compute one
it said3f5a1c9e8b7d2064af13c5e9d78b2a41
truth102723b946d6d281adcf4ba9af3e7529
gemini-3.1-pro · Base64 with unicode — held for 264 characters, then flipped one byte (4oaS4oKs)
it said…IGZpbmFsOiDigKw1My44MzsgcmVjZWlwdCD…
truth…IGZpbmFsOiDigqw1My44MzsgcmVjZWlwdCD…
gpt-5.5 · the 8641st prime — confident, specific, and 76 short
it said89137
truth89213
claude-opus-5 · counting "th" in a 60-word passage — off by one
it said43
truth44
claude-haiku-4.5 · Lord Howe Island → St. John's, both on half-hour offsets — an hour adrift
it said2025-10-04 12:30
truth2025-10-04 13:30

Every task

How many of the 8 models solved each task — unaided vs. with TinyFn tools.

TaskCategoryUnaidedWith TinyFn
md5 hash 0/8 8/8
sha256 hash 0/8 8/8
hmac-sha256 hash 0/8 8/8
crc32 hash 1/8 8/8
nth-prime number 1/8 8/8
rot13 encode 1/8 7/8
base64-encode encode 3/8 8/8
collatz-length number 5/8 8/8
count-th string 5/8 8/8
sample-stddev stats 5/8 8/8
haversine geo 5/8 8/8
base64-decode encode 5/8 7/8
regex-double-letters regex 5/8 7/8
unix-weekday datetime 6/8 8/8
iso-to-unix datetime 6/8 8/8
tz-half-hour-dst time 6/8 8/8
miles-to-km convert 6/8 8/8
liters-to-uk-gallons convert 6/8 8/8
dec-to-hex convert 6/8 8/8
big-multiply math 6/8 8/8
gcd math 6/8 8/8
invert-then-hsl color 6/8 8/8
pearson-correlation stats 6/8 8/8
geometric-mean stats 6/8 8/8
loan-payment finance 6/8 8/8
compound-daily finance 6/8 8/8
days-between-century datetime 7/8 8/8
iso-week-edge datetime 7/8 8/8
add-1000-business-days time 7/8 8/8
hex-encode encode 7/8 8/8
factorial-47 math 7/8 8/8
contrast-ratio color 7/8 8/8
bmr-tdee health 7/8 8/8
add-10000-days datetime 8/8 8/8
weekday-1834 datetime 8/8 8/8
tz-kathmandu-chatham time 8/8 8/8
url-encode encode 8/8 8/8
luhn-19-digit validate 8/8 8/8
percentile-37-5 stats 8/8 8/8
broadcast-address ip 8/8 8/8

Method

Design. Every model runs every task twice with identical prompts: unaided (no tools) and with TinyFn (the task's TinyFn MCP category server, e.g. api.tinyfn.io/mcp/hash/, attached as a toolset via pydantic-ai, models served through OpenRouter).

Grading. Ground truth is computed with the Python standard library, independently of the TinyFn API. Checkers normalize case, whitespace, and number formats, with tolerances on rounded values, and answers must arrive on an explicit ANSWER: line — so verbose models aren't penalized and lucky prose doesn't count. Tool use is verified from message history.

Full transparency. The harness, all 40 task definitions, raw per-evaluation results, and instructions to reproduce are on GitHub: github.com/tinyfn-io/tinybench.

The benchmark keeps score on both sides of the API. Building it exposed three real bugs in TinyFn itself — time/convert silently treating IANA timezone names as UTC, math/multiply losing precision above 253, and four length/speed endpoints using rounded conversion factors. All three are fixed and deployed; the tasks that caught them are still in the suite.

The three remaining tool-condition misses are reported as-is: claude-sonnet-5's ROT13 run hit a provider content filter (high-entropy output, not a tool failure), and claude-haiku-4.5 mis-transcribed one character of a correct Base64 result and miscounted a correct regex result. Nothing is excluded from the totals.

Give your agent the tools

One MCP URL. 500+ deterministic tools. No more plausible-looking hashes.