Multi Step Mathematical Reasoning Evaluation

via “solution step extraction and intermediate reasoning evaluation”

12.5K competition math problems — AMC/AIME/Olympiad level, 7 subjects, standard math benchmark.

Unique: Preserves solution steps as first-class data throughout the evaluation pipeline, enabling evaluation of intermediate reasoning quality rather than just final answers. This supports emerging research on chain-of-thought prompting and interpretable AI reasoning.

vs others: More comprehensive than final-answer-only evaluation because it assesses reasoning quality and interpretability, but requires more manual annotation and is harder to automate than simple answer verification.

via “arithmetic and mathematical reasoning evaluation”

23 hardest BIG-Bench tasks where models initially failed.

Unique: Focuses specifically on multi-step arithmetic and mathematical reasoning through few-shot examples, isolating numerical reasoning capability from general language understanding. Tasks test both calculation accuracy and mathematical inference patterns.

vs others: More focused on mathematical reasoning than general reasoning benchmarks; more accessible than formal mathematics verification because it uses natural language problem statements rather than symbolic notation.

via “multi-step mathematical reasoning benchmark evaluation”

8.5K grade school math problems — multi-step reasoning, verifiable solutions, reasoning benchmark.

Unique: Uses linguistically diverse, human-authored grade school problems (not synthetic) that require genuine multi-step reasoning with basic arithmetic, combined with a standardized answer extraction format (#### delimiter) that enables reproducible evaluation across heterogeneous model outputs

vs others: More challenging than simple arithmetic benchmarks (requires 2-8 reasoning steps) yet more accessible than advanced math benchmarks, making it ideal for measuring practical reasoning improvements in production models

via “reasoning and step-by-step problem decomposition”

text-generation model by undefined. 95,66,721 downloads.

Unique: Emergent chain-of-thought capability from instruction tuning on reasoning datasets; no explicit reasoning module or symbolic engine — reasoning emerges from learned token prediction patterns that favor intermediate explanation tokens, making it lightweight but probabilistic

vs others: Provides transparent reasoning comparable to GPT-4 on simple problems but with full local control; outperforms Mistral-7B on reasoning tasks due to instruction tuning, but lacks the formal verification and symbolic reasoning of specialized tools like Wolfram Alpha

via “mathematical reasoning and step-by-step problem solving”

DeepSeek's 236B MoE model specialized for code.

Unique: Trained on 6 trillion tokens including mathematical reasoning datasets and code-based solutions, enabling both symbolic reasoning and code generation for mathematical problems in a single model without separate math-specific components

vs others: Provides integrated mathematical reasoning and code generation (unlike Copilot which focuses on code) while maintaining open-source weights and supporting local deployment

via “mathematical reasoning with 96.8% gsm8k accuracy”

Largest open-weight model at 405B parameters.

Unique: 405B parameter scale enables 96.8% GSM8K performance through learned chain-of-thought patterns in transformer architecture, achieving near-human accuracy on grade-school math without external symbolic engines or calculators

vs others: Larger model scale than most open-source alternatives improves mathematical reasoning accuracy; however, lacks symbolic verification that specialized math engines provide, making it suitable for reasoning tasks but not formal proofs

via “mathematical reasoning and problem-solving”

671B MoE model matching GPT-4o at fraction of training cost.

Unique: Achieves 90.2% on MATH benchmark through MoE architecture that routes mathematical reasoning tokens through specialized expert parameters, enabling efficient scaling of reasoning capability without proportional increase in active parameters per token

vs others: Matches GPT-4o mathematical reasoning performance (90.2% MATH) while using 37B active parameters vs GPT-4o's undisclosed parameter count, reducing inference latency and cost for math-heavy workloads

via “mathematical reasoning and step-by-step problem solving”

text-generation model by undefined. 1,37,84,608 downloads.

Unique: Qwen2.5-7B-Instruct includes explicit training on mathematical reasoning datasets (including GSM8K, MATH, and proprietary datasets) with emphasis on showing intermediate steps and justifying answers. The instruction-tuning includes prompts that encourage the model to 'think step by step' and 'show your work', which are known to improve mathematical reasoning through in-context learning effects.

vs others: Outperforms base Qwen2.5-7B on mathematical reasoning benchmarks by 15-20% due to instruction-tuning; more accessible than specialized math models (like Minerva) for general-purpose deployment

via “mathematical problem solving with symbolic reasoning”

Cost-efficient reasoning model with configurable effort levels.

Unique: Implements specialized mathematical reasoning patterns with step-by-step derivation generation, achieving competition-level math performance through domain-specific training rather than general reasoning

vs others: Matches o3 on mathematical benchmarks at lower cost; outperforms standard LLMs (GPT-4, Claude) on competition-level problems due to reasoning-grade capabilities

via “multi-step mathematical reasoning evaluation”

Grade school math problems requiring multi-step reasoning

Unique: GSM8K is specifically curated to include a diverse set of multi-step reasoning problems, making it more targeted than generic math datasets, allowing for precise evaluation of reasoning capabilities in LLMs.

vs others: More focused on multi-step reasoning than other benchmarks like MATH, which may include less structured problems.

via “mathematical reasoning and logic problem evaluation with specialized scoring”

ReLE评测：中文AI大模型能力评测（持续更新）：目前已囊括374个大模型，覆盖chatgpt、gpt-5.4、谷歌gemini-3.1-pro、Claude-4.6、文心ERNIE-X1.1、ERNIE-5.0、qwen3.6-max、qwen3.6-plus、百川、讯飞星火、商汤senseChat等商用模型， 以及step3.5-flash、kimi-k2.6、ernie4.5、MiniMax-M2.7、deepseek-v4、Qwen3.6、llama4、智谱GLM-5.1、MiMo-V2、LongCat、gemma4、mistral等开源大模型。不仅提供排行榜，也提供规模超200万的大

Unique: Evaluates mathematical reasoning with 1-5 quality scale for reasoning steps rather than binary correctness, enabling partial credit for correct methodology with computational errors. Combines final answer accuracy with reasoning quality assessment to capture mathematical thinking capability. Includes multi-step reasoning problems and logical inference tasks beyond simple arithmetic.

vs others: More nuanced mathematical assessment than MMLU (binary correctness) and captures reasoning quality vs answer-only evaluation

via “reasoning and step-by-step problem decomposition”

Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors. This 70B instruct-tuned version is optimized for high quality dialogue usecases. It has demonstrated strong...

Unique: Instruction-tuned on datasets containing explicit reasoning traces (e.g., math solutions with working, logic puzzles with step-by-step explanations), enabling the model to learn to generate intermediate reasoning as a learned behavior rather than relying on prompt engineering alone.

vs others: More reliable than base models at producing coherent reasoning chains; comparable to GPT-4 on standard benchmarks but with lower latency and cost, though may underperform on novel reasoning patterns not well-represented in training data.

via “mathematical-problem-solving-with-step-by-step-reasoning”

DeepSeek-V3.1 is a large hybrid reasoning model (671B parameters, 37B active) that supports both thinking and non-thinking modes via prompt templates. It extends the DeepSeek-V3 base with a two-phase long-context...

Unique: Implements explicit reasoning phase specifically optimized for mathematical decomposition, allowing the model to verify intermediate steps before producing final answers, rather than generating answers directly.

vs others: More reliable for complex math than GPT-4 due to explicit verification phase, and more transparent than o1 (which hides reasoning) by allowing users to request step-by-step explanations.

via “complex reasoning and chain-of-thought decomposition”

Command R7B (12-2024) is a small, fast update of the Command R+ model, delivered in December 2024. It excels at RAG, tool use, agents, and similar tasks requiring complex reasoning...

Unique: Command R7B's reasoning is optimized for RAG and tool-use contexts, where intermediate steps can reference retrieved documents or tool outputs, enabling grounded reasoning that combines external knowledge with logical inference

vs others: Outperforms GPT-4 on MATH and AIME benchmarks when combined with tool use for calculation, because it can delegate computation to tools rather than attempting symbolic math in-context

via “mathematical reasoning and symbolic computation”

GLM 4 32B is a cost-effective foundation language model. It can efficiently perform complex tasks and has significantly enhanced capabilities in tool use, online search, and code-related intelligent tasks. It...

Unique: GLM 4 32B includes specialized training on mathematical reasoning datasets, enabling it to show work and explain reasoning — not just generate answers — which is critical for educational and verification use cases

vs others: More cost-effective than Wolfram Alpha for symbolic reasoning while providing better explanations than calculators, though less precise than dedicated symbolic engines for complex expressions

via “mathematical reasoning and symbolic computation”

This is Mistral AI's flagship model, Mistral Large 2 (version mistral-large-2407). It's a proprietary weights-available model and excels at reasoning, code, JSON, chat, and more. Read the launch announcement [here](https://mistral.ai/news/mistral-large-2407/)....

Unique: Trained on mathematical datasets with chain-of-thought reasoning to prioritize step-by-step problem solving, using attention mechanisms that track variable relationships and equation transformations

vs others: Comparable to GPT-4 on mathematical reasoning, while maintaining lower cost; outperforms Llama 2 on complex multi-step problems due to larger parameter count and specialized training

via “mathematical reasoning evaluation”

UGI-Leaderboard — AI demo on HuggingFace

Unique: Isolates mathematical reasoning as a distinct evaluation dimension on the leaderboard, enabling models to be ranked separately on math vs general generation, revealing capability specialization.

vs others: Simpler than running MATH or GSM8K locally with custom evaluation scripts, but less transparent than open-source math benchmarks regarding problem selection and difficulty.

via “reasoning and chain-of-thought task decomposition”

Step 3.5 Flash is StepFun's most capable open-source foundation model. Built on a sparse Mixture of Experts (MoE) architecture, it selectively activates only 11B of its 196B parameters per token....

Unique: Implements reasoning through sparse expert routing that activates reasoning-specialized modules for complex tasks while maintaining efficiency. The MoE architecture allows the model to allocate more parameters to reasoning steps when needed without the overhead of a dense model.

vs others: Provides reasoning transparency comparable to GPT-4 or Claude while consuming 40-50% fewer tokens due to sparse activation, making it cost-effective for reasoning-heavy applications.

via “reasoning and multi-step problem solving”

A 12B parameter model with a 128k token context length built by Mistral in collaboration with NVIDIA. The model is multilingual, supporting English, French, German, Spanish, Italian, Portuguese, Chinese, Japanese,...

Unique: Mistral Nemo's instruction-tuning includes reasoning tasks and chain-of-thought examples, enabling it to generate explicit reasoning steps when prompted. The 128k context window enables longer reasoning chains than smaller-context models.

vs others: Reasoning capability is weaker than larger models (70B+) but sufficient for many reasoning tasks. Prompt-based chain-of-thought is more transparent than implicit reasoning but less efficient than specialized reasoning architectures.

via “reasoning and step-by-step problem decomposition”

Gemma 4 26B A4B IT is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind. Despite 25.2B total parameters, only 3.8B activate per token during inference — delivering near-31B quality at...

Unique: MoE expert specialization enables dedicated reasoning experts that activate for complex reasoning tasks, while general-purpose experts handle simpler steps, optimizing compute allocation across reasoning complexity

vs others: Provides faster reasoning than Llama 3.1 8B (15-20% speedup) while maintaining comparable accuracy on grade-school math and logic puzzles, though underperforms specialized reasoning models like o1-mini on competition-level problems