Early preview
G

Google

Gemini 1.5 Pro

Announced Feb 15, 2024

Much longer inputs become possible

01

What could it do?

Process long collections of text and media within the supported context window.

02

What changed?

Google previewed a context window of up to one million tokens for selected developers and enterprise customers.

WHY IT MATTERED

Long context

Longer inputs made it possible to explore much larger documents or media collections in one request.

03

Where it fell short

A larger input allowance is not a guarantee that every detail will be retrieved or interpreted correctly.

Which release does this page cover?

February 2024 announcement and limited preview, not general availability of every context length. Check sources

Release stage
Limited preview
Highlighted capability
Long context
Preview ceiling
Up to 1 million tokens

What this meant in practice

Context length measures input capacity. Accuracy measures whether the model uses that input correctly. Keeping those concepts separate prevents a larger context window from being mistaken for a proportional increase in intelligence.

ILLUSTRATIVE TASK · NOT A TEST RESULT

Find evidence in a long document

Ask for the passage supporting an answer and its location. Check that quotation against the document. The useful output is an answer you can trace, not just a confident summary.

Common question

Does a million tokens mean a million words?

No. Tokens are pieces of text or other encoded input. Their relationship to words depends on the content and tokenizer.

Sources checked Oct 7, 2026

Release facts were checked against the sources below. Performance claims belong to the developers; we have not independently tested these models. The million token context window was a limited preview at announcement.

Google: Introducing Gemini 1.5 Gemini team: Gemini 1.5 technical report

Benchmark results

EPOCH AI CAPABILITY ESTIMATE
126.9index points
Tested variant: Gemini 1.5 Pro (May 2024)Source interval: 121.1 to 129.2Variant date in source: May 14, 2024

A benchmark estimate, not a percentage or capability multiplier. Reasoning settings are not specified in this source table. Historical estimates can change in later snapshots.

Epoch AI methodology ↗Download the source snapshotChecked Oct 7, 2026 · CC BY 4.0
PUBLISHED BENCHMARK RESULT
1324rating points
Tested: gemini-1.5-pro-001Text Arena Overall · October 8, 2026Reported interval: 1320 to 132879,138 votes

One explicitly named variant per release. Scores come from the same Overall snapshot; preliminary entries and reported intervals are preserved. These are current ratings of earlier variants, not their launch day ratings.

Source: Text Arena ↗Download selected resultsChecked Oct 8, 2026
FOLLOW WHAT HAPPENS NEXT

Breakthroughs, with the followup.

A weekly brief on new discoveries, meaningful checks and what you can actually use.