Benchmark

LVBench

A benchmark for evaluating AI models on long-video understanding tasks, including capabilities such as fine-grained detail recognition and cross-temporal information association within long video content.

Core Facts

Gemini 4 Argon scored 91% on LV Bench (video/long-content understanding), setting a new record.

Verified70% confidence4 sources

Argon achieved a score of 77.9% on the DeepSWE benchmark, setting a new record

Verified65% confidence3 sources

Timeline (last 90 days)

Oct 1

Gemini 4 Argon scored 91% on LV Bench (video/long-content understanding), setting a new record.

Verified70%

Oct 1

Argon achieved a score of 77.9% on the DeepSWE benchmark, setting a new record

Verified65%

Sep 30

Argon set a new best score of 91.7% on LVBench, which measures long video understanding capability.

Unverified50%

All Facts (3)

Verified

Gemini 4 Argon scored 91% on LV Bench (video/long-content understanding), setting a new record.

70%Verified

Argon achieved a score of 77.9% on the DeepSWE benchmark, setting a new record

65%Unverified

Argon set a new best score of 91.7% on LVBench, which measures long video understanding capability.

50%

Source Articles

Gemini 4 Argon Benchmark Analysis: How It Stacks Up Against GPT-6 and ClaudeOct 14 Google Gemini 4 Argon Launches: Million-Token Output Reshapes the Flagship AI LandscapeOct 11 Gemini 4 Argon Launch: Google Reclaims Top-Tier AI Status with Multi-Benchmark WinsOct 1