Research · Benchmark profile

LABBench2

A biology real-world research-task evaluation run with a Linux terminal, scientific tools and restricted internet access.

Reference2Active4 models

Data verified 13 Aug 2026 · Methodology 2.2.0

Benchmark score on LABBench2

Leaderboard · 4 models

Gemini 3.7 Flash leads at 82.1%, followed by GPT-5.6 Terra (81.2%) and Claude Sonnet 5 (80.1%).

Display onlyResearch2

#1Closed

Gemini 3.7 Flash

Google

gemini-3.7-flash

82.1%

Overall #6 · Context 1.05M tokens

#2Closed

GPT-5.6 Terra

OpenAI

gpt-5.6-terra

81.2%

Overall #9 · Context 1.05M tokens

#3Closed

Claude Sonnet 5

Anthropic

claude-sonnet-5

80.1%

Overall #14 · Context 1M tokens

ViewModelsColours

GoogleOpenAIAnthropic

Lumina

LABBench2 leaderboard

Accessible data for LABBench2 leaderboard
Model	Rank	Model ID	Provider	Score	Unit
Gemini 3.7 Flash	#1	gemini-3-7-flash	Google	82.1	percent
GPT-5.6 Terra	#2	gpt-5-6-terra	OpenAI	81.2	percent
Claude Sonnet 5	#3	claude-sonnet-5	Anthropic	80.1	percent
Gemini 3.6 Flash	#4	gemini-3-6-flash	Google	76.1	percent

Benchmark score table (4 models)

One best compatible score per canonical product · higher is better

LABBench2 leaderboard ranked by score
Rank	Model	Provider	License	Evidence use	Score
#1	Gemini 3.7 Flash gemini-3.7-flash	Google	closed	Reference only	82.1%
#2	GPT-5.6 Terra gpt-5.6-terra	OpenAI	closed	Reference only	81.2%
#3	Claude Sonnet 5 claude-sonnet-5	Anthropic	closed	Reference only	80.1%
#4	Gemini 3.6 Flash gemini-3.6-flash	Google	closed	Reference only	76.1%

The top of this snapshot is led by Gemini 3.7 Flash at 82.1%; third place is 2.0 points behind. The top-4 spread is 6.0 points.

Configuration-preserving evidencePublished source rows (4)Exact source labels, configurations, versions and dated observations. Same-snapshot canonical/configuration projections are shown once.

Every distinct dated source-published row for LABBench2, including exact model configuration labels
Exact published model / configuration	Canonical product	Score	Version / system	Evidence	Source / observed
Gemini 3.7 Flash gemini-3-7-flash-medium	Gemini 3.7 Flash	82.1%	2LABBench2 Linux terminal with bioinformatics tools, Python, R and internet	Reference only provider-reported	Google DeepMind permanent refresh source ↗Observed 2026-08-13 · checked 2026-08-13
GPT-5.6 Terra gpt-5-6-terra-max	GPT-5.6 Terra	81.2%	2LABBench2 Linux terminal with bioinformatics tools, Python, R and internet	Reference only provider-reported	Google DeepMind permanent refresh source ↗Observed 2026-08-13 · checked 2026-08-13
Claude Sonnet 5 claude-sonnet-5-max	Claude Sonnet 5	80.1%	2LABBench2 Linux terminal with bioinformatics tools, Python, R and internet	Reference only provider-reported	Google DeepMind permanent refresh source ↗Observed 2026-08-13 · checked 2026-08-13
Gemini 3.6 Flash gemini-3-6-flash-high	Gemini 3.6 Flash	76.1%	2LABBench2 Linux terminal with bioinformatics tools, Python, R and internet	Reference only provider-reported	Google DeepMind permanent refresh source ↗Observed 2026-08-13 · checked 2026-08-13

Compare top models

Gemini 3.7 Flash vs GPT-5.6 Terra GPT-5.6 Terra vs Claude Sonnet 5 Claude Sonnet 5 vs Gemini 3.6 Flash

About LABBench2

Definition and scoring

Organisation: LABBench
Category: Research
Version: 2
Direction: higher is better
Ranking use: Reference
Contamination risk: Unknown

A biology real-world research-task evaluation run with a Linux terminal, scientific tools and restricted internet access. Every genuine source row stays tied to its exact model label, configuration, benchmark version and evaluation system. The summary chart shows one best compatible score per canonical product; Score 2.0 admits only explicitly mapped, frozen protocols and keeps incompatible configurations separate.

Open benchmark source ↗

FAQ

Quick answers

What does LABBench2 measure?

A biology real-world research-task evaluation run with a Linux terminal, scientific tools and restricted internet access.

Which model scores highest on LABBench2?

Gemini 3.7 Flash by Google currently leads with 82.1%.

How many models are evaluated on LABBench2?

4 canonical model products have a qualifying score, backed by 4 distinct dated source rows.

Does this affect overall Lumina rank?

No. This benchmark is display-only and does not enter the overall Lumina composite.

Research · Benchmark profile

LABBench2

A biology real-world research-task evaluation run with a Linux terminal, scientific tools and restricted internet access.

Reference2Active4 models

Data verified 13 Aug 2026 · Methodology 2.2.0

Benchmark score on LABBench2