FrontierCode 2.0
A benchmark for measuring code generation and comprehension ability, consisting of programming tasks with reference answers or executable test cases, scored by task pass rate.
Core Facts
FrontierCode is an evaluation benchmark for AI models' programming capabilities focusing on real engineering tasks rather than simple algorithm problems
FrontierCode 1.1's test cases are typically derived from real open-source projects and require models to have cross-file context understanding and long-sequence reasoning capabilities.
Timeline (last 90 days)
Gemini 4 Carbon scores 40.0% on the coding benchmark FrontierCode 2.0.
An absolute score of 4.0% indicates the model can fully solve about forty percent of the test tasks.
There is currently no clear information on Mythos 5.2's pricing or its benchmark performance on FrontierCode 2.0.
Claude Fable 5.2 scored 49.1% on the FrontierCode 2.0 benchmark, ranking first.
Gemini 4 Argon achieved a score of 60.5% on the code competition benchmark FrontierCode 1.1.
FrontierCode 1.1's test cases are typically derived from real open-source projects and require models to have cross-file context understanding and long-sequence reasoning capabilities.
All Facts (6)
FrontierCode is an evaluation benchmark for AI models' programming capabilities focusing on real engineering tasks rather than simple algorithm problems
80%VerifiedFrontierCode 1.1's test cases are typically derived from real open-source projects and require models to have cross-file context understanding and long-sequence reasoning capabilities.
65%UnverifiedGemini 4 Carbon scores 40.0% on the coding benchmark FrontierCode 2.0.
50%UnverifiedAn absolute score of 4.0% indicates the model can fully solve about forty percent of the test tasks.
50%UnverifiedThere is currently no clear information on Mythos 5.2's pricing or its benchmark performance on FrontierCode 2.0.
50%UnverifiedClaude Fable 5.2 scored 49.1% on the FrontierCode 2.0 benchmark, ranking first.
50%