Share your thoughts, 1 month free Claude Pro on usSee more

Mathematics on AIME 2025 (Accuracy, Training Time)

76.7Accuracy

Document RAG (Google)

Updated 11d ago

Evaluation Results

Method
Document RAG (Google) 2026.04	76.7	-
Reasoning Memory 2026.04	76.7	-
Reasoning Memory 2026.04	75.4	-
Reasoning Memory 2026.04	67.9	-
Reasoning Memory 2026.04	64.6	-
Qwen 3 VL 32B Instruct 2025.12	64.2	-
Length Scaling 2026.04	61.9	-
Length Scaling 2026.04	59.1	-
No RAG 2026.04	58.6	-
Olmo 3.1 32B Instruct 2025.12	57.9	-
Length Scaling 2026.04	52.8	-
Length Scaling 2026.04	50.6	-
Template RAG (ReasonFlux) 2026.04	50	-
Template RAG (Self-Discover) 2026.04	48.3	-
Trajectory RAG (Summary) 2026.04	42.9	-
Document RAG (Google) 2026.04	42.2	-
Zero-shot 2026.04	41.71	-
Trajectory RAG (Prefix) 2026.04	40.8	-
Cog-DRIFT 2026.04	40.28	-
Reasoning Memory 2026.04	39.2	-
No RAG 2026.04	39.2	-
Template RAG (Self-Discover) 2026.04	39.2	-
NuRL (Prefix) 2026.04	38.81	-
Document RAG (CompactDS) 2026.04	38.8	-
Few-shot 2026.04	37.98	-
NuRL (Abstract) 2026.04	36.44	-
Reasoning Memory 2026.04	36.4	-
Length Scaling 2026.04	35.8	-
Trajectory RAG (Prefix) 2026.04	35.8	-
Trajectory RAG (Summary) 2026.04	35.4	-
Template RAG (ReasonFlux) 2026.04	35.4	-
Document RAG (Google) 2026.04	35.1	-
No RAG 2026.04	35	-
RFT 2026.04	34.88	-
Template RAG (ReasonFlux) 2026.04	34.2	-
Document RAG (CompactDS) 2026.04	33.3	-
Length Scaling 2026.04	32.5	-
Trajectory RAG (Summary) 2026.04	32.1	-
Document RAG (CompactDS) 2026.04	31.2	-
Template RAG (Self-Discover) 2026.04	29.6	-
GRPO 2026.04	27.29	-
Trajectory RAG (Prefix) 2026.04	24.6	-
Olmo 3.1 32B Instruct 2025.12	23.3	-
Gemma 3 27B 2025.12	22.9	-
Qwen 3 32B 2025.12	21.3	-
Qwen 2.5 32B 2025.12	13.4	-
Olmo 3.1 32B Instruct 2025.12	8.2	-
NuRL (Abstract) 2026.04	1.11	-
Gemma 2 27B 2025.12	0.9	-
OLMo 2 32B 2025.12	0.9	-
Latent-GRPO 2026.01	0.667	3,108.44
Base 2026.01	0.656	-
GRPO (LLM-Judge) 2026.01	0.633	6,753.47
Zero-shot 2026.04	0.42	-
GRPO 2026.04	0.42	-
NuRL (Prefix) 2026.04	0.42	-
Cog-DRIFT 2026.04	0.42	-
Base 2026.01	0.368	-
Latent-GRPO 2026.01	0.354	2,340.05
GRPO (LLM-Judge) 2026.01	0.338	4,988.84
Few-shot 2026.04	0.33	-
Latent-GRPO 2026.01	0.192	2,280.52
Base 2026.01	0.151	-
GRPO (LLM-Judge) 2026.01	0.117	4,082.96
Apertus 70B 2025.12	0.1	-
RFT 2026.04	0	-