The AI arena is free today

Open Superagent
All benchmarks

Meta Internal Coding Bench

Implementation

Progress Over Time

Interactive timeline showing model performance evolution on Meta Internal Coding Bench

State-of-the-art frontier
Open
Proprietary

Meta Internal Coding Bench Leaderboard

1 models
ContextCostLicense
1
Meta
Muse Spark 1.2
Meta
—1.0M$0.10 / $0.20
Notice missing or incorrect data?
About this benchmark

What is Meta Internal Coding Bench?

Meta's internal evaluation of coding-agent performance.

Meta Internal Coding Bench is a text benchmark evaluating models on agents and code tasks. LLM Stats tracks 1 models on this benchmark, scored on a 0–1 scale. The current average is 0.7, with the leader at 0.7.

Compare leaders on the best AI for agents and best AI for code leaderboards.

Current leaders

Muse Spark 1.2 from Meta currently leads the Meta Internal Coding Bench leaderboard with a score of 0.706 across 1 evaluated AI models.

1Muse Spark 1.2Meta70.6%

FAQ

Common questions about the Meta Internal Coding Bench benchmark and leaderboard.

What is the Meta Internal Coding Bench benchmark?

Meta's internal evaluation of coding-agent performance.

What is the Meta Internal Coding Bench leaderboard?

The Meta Internal Coding Bench leaderboard ranks 1 AI models based on their performance on this benchmark. Currently, Muse Spark 1.2 by Meta leads with a score of 0.706. The average score across all models is 0.706.

What is the highest Meta Internal Coding Bench score?

The highest Meta Internal Coding Bench score is 0.706, achieved by Muse Spark 1.2 from Meta.

How many models are evaluated on Meta Internal Coding Bench?

1 models have been evaluated on the Meta Internal Coding Bench benchmark, with 0 verified results and 1 self-reported results.

Where can I find the Meta Internal Coding Bench dataset?

The Meta Internal Coding Bench dataset is available at https://developer.meta.com/ai/resources/blog/build-with-muse-code/.

What categories does Meta Internal Coding Bench cover?

Meta Internal Coding Bench is categorized under agents and code. The benchmark evaluates text models.

Which model offers the best value on Meta Internal Coding Bench?

Among models scoring within 10% of the leader, Muse Spark 1.2 from Meta is the cheapest, at $0.10 per million input tokens with a score of 0.706.

How recent are the Meta Internal Coding Bench leaderboard results?

The Meta Internal Coding Bench leaderboard was last updated in August 2026 and currently includes 1 evaluated models.
Meta Internal Coding Bench Leaderboard