5 models across 1 family · Latest: MetaMath 70B (2023-10)
AI models focused on mathematics and proofs
MetaMath's portfolio covers 5 active models across 1 non-obsolete family, with task labels spanning general LLM work. Open a model detail page to compare provider routes and sourced benchmarks.
Portfolio context: 0 decision-task tags, 5 active tracked models, latest research stamp 2026-05-19.
Use this portfolio page for
- Teams evaluating general LLM work across this lab's releases
- Readers comparing families before locking a flagship SKU
- 5 tracked SKUs for migration and pricing follow-ups
Do not stop here for
- Choosing a hosting provider without opening a model page for price ladders
Active models
5
Non-deprecated SKUs linked to this researcher
Active families
1
Non-obsolete families in coverage
Open catalog
0 OSS
0 open-weight (text match)
Decision task tags
0
Need benchmarks or capability flags
Latest dated release
2023-10-27
MetaMath 70B
Freshness
2026-05-19
Researched 16d ago
Release cadence
Showing 5 recent dated ships (full timeline below). Latest spotlight: MetaMath 70B (2023-10-27).
Where this lab wins
Task positioning unavailable until capability tags or benchmarks populate for active SKUs.
Flagship quality / price signal
Anchor SKU: MetaMath 70B (best sourced coding Q/$ in this portfolio).
Quality / dollar unavailable for this anchor — missing benchmark coverage and/or output token price on the cheapest ladder route (open the model detail after pricing lands).
MetaMath is an AI research lab founded in N/A. AI models focused on mathematics and proofs. MetaMath ships 1 model family totaling 5 models, with the most recent release MetaMath 70B in 2023-10. Notable families include MetaMath. Use it as a stable reference for lab background, release coverage, and follow-up model pages as they are added. Researchers and evaluators can scan counts. View official API endpoints, benchmark performance, and coding/agent fit for every MetaMath model.
About
MetaMath stands at the forefront of artificial intelligence research, particularly in the realm of enhancing large language models' (LLMs) mathematical reasoning capabilities. By developing a unique methodology, MetaMath has addressed the limitations of existing LLMs in solving complex mathematical problems, a domain that has traditionally presented significant challenges. Through its pioneering approach, MetaMath has substantially bridged this gap, positioning itself as a leader in this specialized field. Central to MetaMath's success is its innovative bootstrapping technique, which involves rewriting mathematical questions from multiple perspectives to craft the MetaMathQA dataset. This strategy not only enriches the dataset but also ensures a more robust training ground for the LLaMA-2 models. This creates an alternative to conventional training methods that rely heavily on standard datasets, enabling MetaMath models to better understand and solve diverse mathematical problems. The impact of this approach is evident in the striking results achieved by the MetaMath models. With the MetaMath-7B model attaining impressive accuracy on benchmarks like GSM8K and MATH, and the larger MetaMath-70B model outperforming even GPT-3.5-Turbo on GSM8K, the project demonstrates the significant advancements that can be made through targeted data enhancement. Such successes not only highlight the potential of fine-tuned LLMs but also illustrate the importance of high-quality, diverse datasets in pushing the boundaries of AI. Moreover, the public availability of the MetaMathQA dataset and the resulting models underscores MetaMath's commitment to collaboration and transparency in AI research. By sharing these resources, MetaMath promotes further exploration and innovation in the field, inviting researchers worldwide to contribute to its ongoing developments. This openness not only accelerates progress but also inspires new applications of LLMs in various complex domains, reinforcing MetaMath's role as a pivotal contributor to generative AI and language model research.
Featured models
| Model | Released | Context | Input price ($/1M) | Output price ($/1M) | License |
|---|---|---|---|---|---|
| MetaMath 70B | 2023-10-27 | - | - | - | Apache 2.0 |
| MetaMath 13B | 2023-10-27 | - | - | - | Apache 2.0 |
| MetaMath 7B | 2023-10-27 | - | - | - | Apache 2.0 |
Model families
Recent releases
- MetaMath 70B- 2023-10-27
- MetaMath 13B- 2023-10-27
- MetaMath 7B- 2023-10-27
- MetaMath Mistral 7B- 2023-10-27
- MetaMath Llemma 7B- 2023-10-27
FAQ
Who founded MetaMath and when?
MetaMath was founded in N/A and is associated with N/A.
What models has MetaMath released?
MetaMath ships 5 models across 1 family: MetaMath.
Is MetaMath's technology open source?
LLMReference does not yet have enough model license data to classify MetaMath's releases.
Where is MetaMath headquartered?
MetaMath is headquartered in N/A.
What is MetaMath known for?
AI models focused on mathematics and proofs. Its most prominent tracked family is MetaMath.
How can I access MetaMath's models?
MetaMath's provider availability is tracked on model pages as API and hosting data is verified.
Explore related pages
Last reviewed: 2026-05-19. Data sourced from public lab announcements and provider documentation.