The Lab's own votes, not a benchmark and not part of the Derp Score.

Score = mean of the Lab's votes (Smart 0, Meh 50, Dumb 100), all scored by the current grader; lower is smarter. Per model only: nothing is added up across agents.

LOADING COMMUNITY DATA

How derpy is your AI today?

DAILY RUN · MODELS

How each model is changing

One card per model: its score, time and tokens in every daily run on today's chores, with a dashed line where its CLI version changed. Hover a chart (or focus it and use the arrow keys) for each run's numbers. Up is worse on every chart. Lower score is smarter.

Daily runs · last 3 days (UTC)

2 daily runs on the current 10 chores, 2026-11-12 18:30 UTC to 2026-11-13 18:30 UTC. Changes read: ▲ worse, ▼ better, ≈ about the same; “vs previous” is the run before, “vs first” the oldest run here.

Claude Fable 5.2

Claude Code 2.1.323 · 1 runs

Lab score25.0

Time per chore91s

Output tokens per chore8.1k

Reasoning tokens per chore2.5k

Newest run

S 5 · M 5 · D 0

Not Smart, all 1 runs

Too slow5

Every chore, run by run

11-12 11-13

Pantry

impossible-test

·M

trap-missing-file

·S

Larder

larder-fefo

·S

larder-i18n

·M

larder-roles

·M

larder-sales

·M

larder-sites

·M

Python-Markdown

pymd-heading-offset

·S

Run by run (table)

Run (UTC)	Score	S / M / D	Time per chore	Output / reasoning per chore	CLI
2026-11-13 18:30 UTC	25.0	5 / 5 / 0	91s · 5 overlapped	8.1k / 2.5k	2.1.323

Claude Opus 5.5

Claude Code 2.1.323 · 2 runs

Lab score25.0

▲ +5.0 vs previous▲ +5.0 vs first

Time per chore97s

≈ +9% vs previous≈ +9% vs first

Output tokens per chore8.0k

≈ +1% vs previous≈ +1% vs first

Reasoning tokens per chore1.7k

▼ -12% vs previous▼ -12% vs first

CLI changes (dashed lines): 2.1.323 from 11-13 18h.

Newest run

S 5 · M 5 · D 0

Not Smart, all 2 runs

Too slow7

Claude Sonnet 5.5

Claude Code 2.1.323 · 2 runs

Lab score20.0

▲ +15.0 vs previous▲ +15.0 vs first

Time per chore72s

≈ +9% vs previous≈ +9% vs first

Output tokens per chore5.4k

≈ +9% vs previous≈ +9% vs first

Reasoning tokens per chore1.0k

≈ +5% vs previous≈ +5% vs first

CLI changes (dashed lines): 2.1.323 from 11-13 18h.

Newest run

S 6 · M 4 · D 0

Not Smart, all 2 runs

Too slow

GPT-6.1 Sol

OpenAI Codex 0.163.0 · 2 runs

Lab score25.0

▲ +10.0 vs previous▲ +10.0 vs first

Time per chore73s

≈ +4% vs previous≈ +4% vs first

Output tokens per chore2.2k

≈ -5% vs previous≈ -5% vs first

Reasoning tokens per chore98

≈ +8% vs previous≈ +8% vs first

Newest run

S 5 · M 5 · D 0

Not Smart, all 2 runs

Too slow6
Poor output2

Every chore, run by run

GPT-6.1 Terra

OpenAI Codex 0.163.0 · 2 runs

Lab score15.0

≈ 0.0 vs previous≈ 0.0 vs first

Time per chore68s

▲ +22% vs previous▲ +22% vs first

Output tokens per chore2.3k

▲ +13% vs previous▲ +13% vs first

Reasoning tokens per chore73

≈ -4% vs previous≈ -4% vs first

Newest run

S 7 · M 3 · D 0

Not Smart, all 2 runs

Too slow3
Poor output3

Every chore, run by run

Gemini 4 Carbon (medium)

Antigravity 1.5.8 · 2 runs

Lab score80.0

▲ +5.0 vs previous▲ +5.0 vs first

Time per chore518s

≈ +6% vs previous≈ +6% vs first

Output tokens per chore24k

▼ -14% vs previous▼ -14% vs first

Reasoning tokens per chore18k

≈ -9% vs previous≈ -9% vs first

CLI changes (dashed lines): 1.5.8 from 11-13 18h.

Newest run

S 0 · M 4 · D 6

Not Smart, all 2 runs

Repeated mistake

Gemini 4 Argon (medium)

Antigravity 1.5.8 · 2 runs

Lab score70.0

▲ +10.0 vs previous▲ +10.0 vs first

Time per chore462s

≈ +5% vs previous≈ +5% vs first

Output tokens per chore23k

≈ -9% vs previous≈ -9% vs first

Reasoning tokens per chore17k

≈ -5% vs previous≈ -5% vs first

CLI changes (dashed lines): 1.5.8 from 11-13 18h.

Newest run

S 1 · M 4 · D 5

Not Smart, all 2 runs

Gemini 4 Hydrogen (medium)

Antigravity 1.5.8 · 2 runs

Lab score25.0

≈ 0.0 vs previous≈ 0.0 vs first

Time per chore134s

▲ +12% vs previous▲ +12% vs first

Output tokens per chore13k

≈ +0% vs previous≈ +0% vs first

Reasoning tokens per chore7.0k

≈ -1% vs previous≈ -1% vs first

CLI changes (dashed lines): 1.5.8 from 11-13 18h.

Newest run

S 5 · M 5 · D 0

Not Smart, all 2 runs

Too slow7

This week against the week before, per model

Claude Code · plan max · CLI 2.1.323

Model	This week	Week before	Votes	Most common reason	Avg time	Tokens in / out	Quota per run (5h · week)
Claude Fable 5.2	25.0	—	10	Too slow · 5	91s	229k / 8k	1.00 · 0.20 pt
Claude Opus 5.5	18.3	18.6	60	Too slow · 20	93s	398k / 8k	0.58 · 0.06 pt
Claude Sonnet 5.5	6.7	1.4	60	Too slow · 6	62s	250k / 5k	0.41 · 0.00 pt

OpenAI Codex · plan plus · CLI 0.163.0

Model	This week	Week before	Votes	Most common reason	Avg time	Tokens in / out	Quota per run (5h · week)
GPT-6.1 Sol	17.5	9.3	60	Too slow · 15	67s	158k / 2k	1.20 · 0.20 pt

pymd-slug

Run by run (table)

Run (UTC)	Score	S / M / D	Time per chore	Output / reasoning per chore	CLI
2026-11-13 18:30 UTC	25.0	5 / 5 / 0	97s · 6 overlapped	8.0k / 1.7k	2.1.323 (new)
2026-11-12 18:30 UTC	20.0	6 / 4 / 0	89s · 6 overlapped	7.9k / 2.0k	2.1.322

Run by run (table)

Run (UTC)	Score	S / M / D	Time per chore	Output / reasoning per chore	CLI
2026-11-13 18:30 UTC	20.0	6 / 4 / 0	72s · 2 overlapped	5.4k / 1.0k	2.1.323 (new)
2026-11-12 18:30 UTC	5.0	9 / 1 / 0	66s · 8 overlapped	4.9k / 999	2.1.322

Run (UTC)	Score	S / M / D	Time per chore	Output / reasoning per chore	CLI
2026-11-13 18:30 UTC	25.0	5 / 5 / 0	73s	2.2k / 98	0.163.0
2026-11-12 18:30 UTC	15.0	7 / 3 / 0	70s	2.3k / 91	0.163.0

Run (UTC)	Score	S / M / D	Time per chore	Output / reasoning per chore	CLI
2026-11-13 18:30 UTC	15.0	7 / 3 / 0	68s	2.3k / 73	0.163.0
2026-11-12 18:30 UTC	15.0	7 / 3 / 0	56s	2.0k / 76	0.163.0

Run by run (table)

Run (UTC)	Score	S / M / D	Time per chore	Output / reasoning per chore	CLI
2026-11-13 18:30 UTC	80.0	0 / 4 / 6	518s · 2 overlapped	24k / 18k	1.5.8 (new)
2026-11-12 18:30 UTC	75.0	0 / 5 / 5	490s · 2 overlapped	27k / 20k	1.5.5

Run by run (table)

Run (UTC)	Score	S / M / D	Time per chore	Output / reasoning per chore	CLI
2026-11-13 18:30 UTC	70.0	1 / 4 / 5	462s · 2 overlapped	23k / 17k	1.5.8 (new)
2026-11-12 18:30 UTC	60.0	1 / 6 / 3	440s · 2 overlapped	25k / 18k	1.5.5

Run (UTC)	Score	S / M / D	Time per chore	Output / reasoning per chore	CLI
2026-11-13 18:30 UTC	25.0	5 / 5 / 0	134s	13k / 7.0k	1.5.8 (new)
2026-11-12 18:30 UTC	25.0	5 / 5 / 0	120s	13k / 7.1k	1.5.7

GPT-6.1 Terra	11.7	7.9	60	Poor output · 7	59s	177k / 2k	0.75 · 0.10 pt

Antigravity · CLI 1.5.8

Model	This week	Week before	Votes	Most common reason	Avg time	Tokens in / out	Quota per run (5h · week)
Gemini 4 Carbon (medium)	77.5	—	20	Repeated mistake · 11	504s	141k / 26k	0.83 · 0.36 pt
Gemini 4 Argon (medium)	68.3	62.1	60	Too slow · 33	441s	107k / 23k	0.80 · 0.48 pt
Gemini 4 Hydrogen (medium)	25.0	—	20	Too slow · 7	127s	59k / 13k	0.32 · 0.14 pt