Math

Models 7

• Gen 2 Mistral
  Leanstral 1.5 119B A6B

  By Mistral AI

  Leanstral 1.5 119B A6B is a mixture of experts model with 119 billion total parameters and 6.5 billion active per token, built by Mistral AI as a code agent for formal verification in Lean 4. It accepts text and image input and produces text output, with a 256k token context window. It solves competition math benchmarks like miniF2F and PutnamBench and can verify code correctness properties, uncovering real bugs in open source repositories.

  🐞Debugging🤖Agents🔢Math
  NewCoding

  Released 5d ago
• Gen 7 LongCat
  LongCat-2.0

  By Meituan

  LongCat-2.0 is an open-source MoE language model with 1.6 trillion total parameters (~48B activated per token) and a 1M-token context window. It introduces LongCat Sparse Attention for efficient long-context processing and was pretrained on 35+ trillion tokens using AI ASIC hardware. Strong focus on coding, agentic, and reasoning tasks.

  📚Large Language Models💻Coding🤖Agents🔢Math
  NewText

  Released 7d ago
• Gen 3
  ZAYA1 8B

  By Zyphra AI

  ZAYA1-8B is Zyphra’s small mixture-of-experts reasoning model, built for long-form reasoning, mathematics, and coding. It has 8.4B total parameters with 760M active parameters, is post-trained for reasoning, and is positioned as highly intelligence-efficient enough for both strong test-time-compute use and local on-device deployment.

  📚Large Language Models🔎Problem solving💻Coding🔢Math
  NewMultimodal

  Released 2mo ago
• Gen 7
  ERNIE 5.1

  By Baidu

  ERNIE-5.1 is Baidu’s new preview flagship language model, built for stronger general text capability, better cost efficiency, and improved creative performance. It is positioned as the top-ranked Chinese model on the LMArena Text leaderboard and as a high-efficiency upgrade over ERNIE-5.0 with much lower training cost at its scale.

  💬Chatting💼Business💻Coding🤖Agents⚖️Legal🔢Math
  NewText

  Released 2mo ago
• Gen 3
  Qwen 3.6 35B A3B

  By Alibaba

  Qwen3.6-35B-A3B is Qwen’s open-weight multimodal MoE model for coding, agentic workflows, long-context reasoning, and vision-language tasks. It has 35B total parameters with 3B activated, supports image-text-to-text use, preserves reasoning context across turns, and natively handles 262,144 tokens with extension up to about 1.01M.

  💻Coding🤖Agents🔍Image interpretation🔢Math💻Conversational coding
  NewMultimodal

  Released 2mo ago
• Gen 3 Mistral
  Mistral Small 4

  By Mistral AI

  Mistral Small 4 is an open hybrid model that unifies instruct, reasoning, and coding in a single multimodal model with a 256k context window.

  💬Chatting🔍Advanced reasoning🔍Image interpretation🔢Math💻Conversational coding
  Multimodal

  Released 3mo ago
• Gen 7
  Apertus 70B Instruct 2509

  By Swiss AI Initiative

  Apertus 70B Instruct is a fully open 70B-parameter decoder-only language model trained from scratch on 15T tokens using a staged curriculum of web, code, and math data. It natively supports 1811 languages and a 65,536-token context window. Post-trained with SFT and QRPO alignment, it supports tool use and agentic workflows. Weights, training data, and recipes are all publicly released. Licensed Apache 2.0.

  📝Text generation💻Coding🤖Agents🌍Translations🔢Math
  Text

  Released 9mo ago