Benchmark

IFEval

What is IFEval?

Verifiable-constraint subset only; does not measure answer quality. Benchmark Radar tracks 4 reported scores for IFEval, cited by 9 source documents, with…

Released: 2023-11-14
Evaluates: instruction following
Openness: unknown
Importer: Model reports
Reported scores: 4

Source provenance

Original evidence https://github.com/google-research/google-research/tree/master/instruction_following_eval

IFEval code

Code https://github.com/google-research/google-research/tree/master/instruction_following_eval

IFEval results and reported scores

Model reports

Model	Organization	Reported value	Reported	Evidence
DeepSeek-R1	DeepSeek	83.3	2025-01-22	evidence
DeepSeek-V3	DeepSeek	86.1	2024-12-27	evidence
Qwen3-235B-A22B (Thinking)	Qwen	83.4	2025-05-14	evidence
Qwen3.5-397B-A17B	Qwen	92.6	2026-02-16	evidence

Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.

Which sources cite IFEval?

Qwen3.5-397B-A17B Qwen, 2026-02-16
Olmo 3 (7B and 32B) Ai2, 2025-11-20
Qwen3 Qwen, 2025-05-14
Mistral Medium 3 Mistral, 2025-05-07
GPT-4.1 OpenAI, 2025-04-14
Claude 3.7 Sonnet Anthropic, 2025-02-24
DeepSeek-R1 DeepSeek, 2025-01-22
DeepSeek-V3 DeepSeek, 2024-12-27
Llama 3.1 405B Meta, 2024-07-23