Topic hub · 132 claims
LLM releases 2024–2025 — frontier and open-weight catalog
The frontier-model releases that defined 2024 and the first half of 2025. Hand-verified release dates with model cards and official announcements.
The frontier-model release cadence
From 2023 to 2025 the frontier-model release cadence accelerated from quarterly to monthly. Every major lab shipped multiple iterations: OpenAI's GPT-4 → 4-Turbo → 4o → o-series; Anthropic's Claude 3 family → 3.5 family → Opus 4; Google's Gemini Pro → 1.5 Pro → Ultra → 2.0; Meta's Llama 2 → 3 → 3.1 → 3.2 → 3.3. The pace produced two problems: (1) developers couldn't track which model was current; (2) blog posts went stale in weeks. This hub is the canonical reference.
The open-weight wave
2024 was the year open-weight models reached frontier parity for many tasks. Llama 3.1 405B (Meta 2024-07), Mistral Large (Mistral 2024), Qwen 2.5 (Alibaba 2024), DeepSeek V3 (DeepSeek 2024-12), DeepSeek-R1 (DeepSeek 2025-01), Phi-4 (Microsoft 2024-12), Gemma (Google 2024). Each release came with model card, official blog post, and Hugging Face deployment — citable primary sources, not just speculation.
Reasoning models — the new axis
DeepSeek-R1 (January 2025) demonstrated that reinforcement-learning-trained reasoning models could match much larger conventional models on math + code benchmarks. OpenAI's o1 + o3 series brought test-time compute scaling. The trend redefined what "capable model" means: not just parameter count, but inference-time reasoning budget.
Defined terms (3)
- Frontier model
- A language model representing the current state of the art for general-purpose capabilities at its release date.
- Open-weight model
- A model whose trained weights are publicly downloadable. May or may not include training data + training code.
- Reasoning model
- A model trained to produce extended chain-of-thought before its final answer, often with RL-tuned reasoning rewards. DeepSeek-R1, OpenAI o1/o3 are examples.
All claims in this topic (132)
- AI Index Report 2024·publicly released on 2024-04-15 by Stanford HAI — 7th annual report measuring AI trends across research, industry, ethics, education, policy(1.00 · 2 sources)
- AI21 Jamba·publicly released on 2024-03-28 by AI21 Labs — first production-grade hybrid SSM-Transformer (Mamba + Transformer) model, 52B/12B active(1.00 · 2 sources)
- Allen AI Molmo·publicly released on 2024-09-25 by Allen Institute for AI — fully-open multimodal VLM family (1B/7B/72B), Apache 2.0(1.00 · 2 sources)
- Allen AI OLMo 2·publicly released on 2024-11-26 by Allen Institute for AI — fully-open 7B + 13B models with full training data, code, recipes(1.00 · 2 sources)
- AlphaFold 3·released on 2024-05-08(1.00 · 2 sources)
- Anthropic ChatGPT Edu·publicly released on 2024-05-30 by OpenAI — ChatGPT Edu (university-focused tier with GPT-4o + admin controls)(1.00 · 2 sources)
- Anthropic Citations API·publicly released on 2025-01-23 by Anthropic — built-in source citations(1.00 · 2 sources)
- Anthropic Claude Code·publicly released on 2025-02-24 by Anthropic — coding agent CLI(1.00 · 2 sources)
- Anthropic Claude for Education·publicly released on 2025-04-02 by Anthropic — Learning mode + Claude API for education institutions + Northeastern + LSE + Champlain partnerships(1.00 · 2 sources)
- Anthropic Claude Haiku 3.5·released on 2024-11-04 by Anthropic(1.00 · 2 sources)
- Anthropic Claude Opus 4·released on 2025-05-22 by Anthropic(1.00 · 2 sources)
- Anthropic Claude Sonnet 4·released on 2025-05-22 by Anthropic(1.00 · 2 sources)
- Anthropic Computer Use·publicly released on 2024-10-22 by Anthropic — Claude 3.5 Sonnet computer use beta(1.00 · 2 sources)
- Anthropic Constitutional Classifiers·publicly released on 2025-02-04 by Anthropic — safeguard against jailbreaks via constitutional-trained input/output filters(1.00 · 2 sources)
- Anthropic Tool Use (general availability)·publicly released on 2024-05-30 by Anthropic(1.00 · 2 sources)
- Apple Foundation Models·publicly released on 2024-06-10 by Apple — Apple Intelligence foundation models announced at WWDC 2024 (on-device + Private Cloud Compute)(1.00 · 2 sources)
- Apple Intelligence·publicly released on 2024-10-28 by Apple(1.00 · 2 sources)
- Black Forest Labs Flux·publicly released on 2024-08-01 — Flux.1 [pro/dev/schnell] image generation(1.00 · 2 sources)
- Bolt.new·publicly released on 2024-10 by StackBlitz — AI web app builder generating full-stack apps from natural-language prompts(1.00 · 2 sources)
- ChatGPT search·publicly released on 2024-10-31 by OpenAI(1.00 · 2 sources)
- Claude 3 family (Opus, Sonnet, Haiku)·released on 2024-03-04(1.00 · 1 sources)
- Claude 3.5 Haiku·released on 2024-11-04(1.00 · 2 sources)
- Claude 3.5 Sonnet·released on 2024-06-20(1.00 · 1 sources)
- Claude 3.7 Sonnet·publicly released on 2025-02-24 by Anthropic — first hybrid-reasoning model with optional extended-thinking mode(1.00 · 2 sources)
- Cohere Aya 23·publicly released on 2024-05-22 by Cohere For AI — multilingual model covering 23 languages(1.00 · 2 sources)
- Cohere Aya Vision·publicly released on 2025-03-04 by Cohere For AI — multilingual open-weight vision-language models (8B + 32B), 23 languages(1.00 · 2 sources)
- Cohere Command A·publicly released on 2025-03-13 by Cohere — flagship enterprise model with 256k context, agentic workflows focus(1.00 · 2 sources)
- Cohere Command R+·publicly released on 2024-04-04 by Cohere(1.00 · 2 sources)
- Cohere Embed v4·publicly released on 2025-04-09 by Cohere — multimodal embedding model, 256k context support, 100+ languages(1.00 · 2 sources)
- CRAG (Corrective RAG)·introduced in Yan et al. 2024 — corrective retrieval-augmented generation(1.00 · 2 sources)
- Databricks DBRX·publicly released on 2024-03-27 by Databricks — 132B-parameter MoE (36B active per token), Databricks Open Model License(1.00 · 2 sources)
- DeepSeek-R1·released on 2025-01-20 with reasoning chain-of-thought capabilities(1.00 · 2 sources)
- DeepSeek-V2·publicly released on 2024-05-07 by DeepSeek — MoE 236B-parameter open model(1.00 · 2 sources)
- DeepSeek-V3·publicly released on 2024-12-26 by DeepSeek AI — 671B-parameter MoE (37B active), open weights(1.00 · 2 sources)
- ElevenLabs Conversational AI·publicly released on 2024-11-19 by ElevenLabs — voice-to-voice agent platform combining ASR + LLM + TTS in single API(1.00 · 2 sources)
- Gemini 2.5 Pro·publicly released on 2025-03-25 by Google DeepMind(1.00 · 2 sources)
- Gemini Ultra·publicly released on 2024-02-08 via Gemini Advanced subscription(1.00 · 2 sources)
- Gemma·released on 2024-02-21(1.00 · 2 sources)
- GenAI Apps SDK (Anthropic)·publicly released on 2024-09-12 — Message Batches API by Anthropic(1.00 · 2 sources)
- Genmo Mochi 1·publicly released on 2024-10-22 by Genmo — open-weight text-to-video diffusion model, 10B parameters(1.00 · 2 sources)
- Google Gemma 2·publicly released on 2024-06-27 by Google DeepMind — Gemma 2 family (9B + 27B), Gemma terms(1.00 · 2 sources)
- Google Gemma 3·publicly released on 2025-03-12 by Google DeepMind — Gemma 3 family (1B/4B/12B/27B), 128k context, multimodal vision(1.00 · 2 sources)
- Google Veo 2·publicly released on 2024-12-16 by Google DeepMind — 4K text-to-video generation, improved physics + cinematography(1.00 · 2 sources)
- GPT-4o·released on 2024-05-13(1.00 · 1 sources)
- GPT-4o mini·released on 2024-07-18(1.00 · 1 sources)
- GraphRAG·introduced in Edge et al. 2024 — Microsoft Research knowledge-graph RAG(1.00 · 2 sources)
- Grok 3·publicly released on 2025-02-17 by xAI(1.00 · 2 sources)
- Hugging Face SmolLM·publicly released on 2024-07-16 by Hugging Face — small-LM family (135M/360M/1.7B) optimized for on-device inference(1.00 · 2 sources)
- IBM Granite·publicly released on 2024-05-09 by IBM — open-weight enterprise-AI model family (3B/8B/13B/20B/34B), Apache 2.0(1.00 · 2 sources)
- IBM Granite 3·publicly released on 2024-10-21 by IBM — Granite 3 family (2B/8B + MoE variants), Apache 2.0 enterprise focus(1.00 · 2 sources)
- Inception Labs Mercury·publicly released on 2025-02-26 by Inception Labs — first commercial diffusion LLM (dLLM), Mercury Coder Small + Mini(1.00 · 2 sources)
- LangGraph·publicly released on 2024-01-17 by LangChain — agent-graph framework(1.00 · 2 sources)
- Liquid AI LFM·publicly released on 2024-09-30 by Liquid AI — Liquid Foundation Models (LFM-1B/3B/40B) using novel non-Transformer architecture(1.00 · 2 sources)
- Llama 3·released on 2024-04-18(1.00 · 2 sources)
- Llama 3.1·released on 2024-07-23(1.00 · 2 sources)
- Llama 3.2 (multimodal release including 11B and 90B vision models)·released on 2024-09-25(1.00 · 2 sources)
- Llama 3.3 70B·released on 2024-12-06(1.00 · 1 sources)
- Llama 4·released on 2025-04-05 by Meta — Scout + Maverick + Behemoth lineup(1.00 · 2 sources)
- Magentic-One·publicly released on 2024-11-04 by Microsoft Research — generalist multi-agent system built on AutoGen(1.00 · 2 sources)
- Mamba-2·introduced in Dao & Gu 2024 — structured state space duality(1.00 · 2 sources)
- Meta Llama 3.2 Vision·publicly released on 2024-09-25 by Meta — 11B + 90B vision-language variants of Llama 3.2(1.00 · 2 sources)
- Meta Movie Gen·publicly released on 2024-10-04 by Meta AI — research preview of 30B-parameter text-to-video + 13B text-to-audio generation(1.00 · 2 sources)
- Meta SAM 2·publicly released on 2024-07-29 by Meta AI — Segment Anything Model 2, real-time video segmentation(1.00 · 2 sources)
- Microsoft Phi-4 Multimodal·publicly released on 2025-02-26 by Microsoft — Phi-4-multimodal 5.6B parameters with audio + image + text(1.00 · 2 sources)
- Mistral Codestral·publicly released on 2024-05-29 by Mistral AI — code-specialized model(1.00 · 2 sources)
- Mistral Large 2·released on 2024-07-24 by Mistral AI(1.00 · 2 sources)
- Mistral Le Chat·publicly released on 2024-02-26 by Mistral AI — consumer chat assistant interface to Mistral models(1.00 · 2 sources)
- Mistral Nemo·publicly released on 2024-07-18 by Mistral AI + NVIDIA — 12B model with 128k context, Apache 2.0(1.00 · 2 sources)
- Mistral OCR·publicly released on 2025-03-06 by Mistral AI — document-understanding OCR model with high-fidelity table + math extraction(1.00 · 2 sources)
- Mistral Pixtral 12B·publicly released on 2024-09-11 by Mistral AI — 12B multimodal vision-language model, Apache 2.0(1.00 · 2 sources)
- Mistral Saba·publicly released on 2025-02-17 by Mistral AI — 24B model optimized for Arabic + South Asian languages(1.00 · 2 sources)
- Mistral Small 3·publicly released on 2025-01-30 by Mistral AI(1.00 · 2 sources)
- Model Context Protocol (MCP)·publicly released on 2024-11-25 by Anthropic(1.00 · 2 sources)
- MoE Mixtral 8x22B·released on 2024-04-10 by Mistral AI(1.00 · 2 sources)
- NVIDIA Cosmos·publicly released on 2025-01-06 by NVIDIA — World Foundation Models platform for physical AI + robotics + autonomous-vehicle training(1.00 · 2 sources)
- NVIDIA Nemotron-4 340B·publicly released on 2024-06-14 by NVIDIA — 340B-parameter open-weight model optimized for synthetic data generation(1.00 · 2 sources)
- NVIDIA NIM·publicly released on 2024-03-18 by NVIDIA — inference microservices for foundation models(1.00 · 2 sources)
- OLMo·released on 2024-02-01 by Allen Institute for AI(1.00 · 2 sources)
- OpenAI Batch API·publicly released on 2024-04-15 by OpenAI(1.00 · 2 sources)
- OpenAI Codex (2025 cloud agent)·publicly released on 2025-05-16 — software-engineering agent in ChatGPT cloud(1.00 · 2 sources)
- OpenAI o1·publicly released on 2024-12-05 — full version with reasoning chain(1.00 · 2 sources)
- OpenAI o3-mini·publicly released on 2025-01-31 by OpenAI(1.00 · 2 sources)
- OpenAI o4-mini·publicly released on 2025-04-16 by OpenAI — successor to o3-mini reasoning model with multimodal input + tool use(1.00 · 2 sources)
- OpenAI Operator·publicly released on 2025-01-23 by OpenAI — agentic browser-control agent(1.00 · 2 sources)
- OpenAI Realtime API·publicly released on 2024-10-01 by OpenAI(1.00 · 2 sources)
- OpenAI Structured Outputs·publicly released on 2024-08-06 — guaranteed JSON Schema conformance(1.00 · 2 sources)
- Phi-3 (Microsoft small language model family)·released on 2024-04-23(1.00 · 2 sources)
- Phi-4·released on 2024-12-12 by Microsoft Research(1.00 · 2 sources)
- Qwen 2.5·released on 2024-09-19 by Alibaba(1.00 · 2 sources)
- SGLang·introduced in Zheng et al. 2024 — efficient LLM serving with structured outputs(1.00 · 2 sources)
- Snowflake Arctic·publicly released on 2024-04-24 by Snowflake — 480B-parameter MoE LLM (17B active), Apache 2.0(1.00 · 2 sources)
- Sora (public release to ChatGPT Plus subscribers)·released on 2024-12-09(1.00 · 1 sources)
- Stability AI Stable Audio 2.0·publicly released on 2024-04-03 by Stability AI — Stable Audio 2.0 long-form music + sound effects from text prompts(1.00 · 2 sources)
- Stability AI Stable Diffusion 3.5·publicly released on 2024-10-22 by Stability AI — SD 3.5 Large (8B) + Medium + Large Turbo open-weight image generation(1.00 · 2 sources)
- Stability AI Stable Video 4D·publicly released on 2024-07-24 by Stability AI — first model generating dynamic novel-view synthesis (4D = 3D + time) from single video input(1.00 · 2 sources)
- Stable Diffusion 3·publicly released on 2024-06-12 (Stable Diffusion 3 Medium weights)(1.00 · 2 sources)
- Suno v3·publicly released on 2024-03-21 by Suno — improved music generation (full 2-minute songs, lyrics + instrumentation)(1.00 · 2 sources)
- Suno v4·publicly released on 2024-11-19 by Suno — improved music generation, longer outputs, audio remastering(1.00 · 2 sources)
- Tencent Hunyuan Video·publicly released on 2024-12-04 by Tencent — 13B-parameter open-weight text-to-video model(1.00 · 2 sources)
- Tencent Hunyuan-Large·publicly released on 2024-11-05 by Tencent — 389B-parameter open-weight MoE (52B active)(1.00 · 2 sources)
- Tülu 3·publicly released on 2024-11-21 by Allen Institute for AI(1.00 · 2 sources)
- Windsurf·publicly released on 2024-11-13 by Codeium — AI-native IDE editor (renamed from Codeium IDE), forked from VSCode(1.00 · 2 sources)
- xAI Grok-2·publicly released on 2024-08-14 by xAI — Grok-2 + Grok-2 mini release on the X platform(1.00 · 2 sources)
- Black Forest Labs Flux.1 Kontext·publicly released on 2025-05-29 by Black Forest Labs — image generation with in-context editing (text + image input)(0.95 · 2 sources)
- DeepSeek V3·released on 2024-12-26(0.95 · 2 sources)
- Devin·publicly released on 2024-03-12 by Cognition Labs(0.95 · 2 sources)
- Firecrawl·publicly released on 2024-04-15 by Mendable AI — web-scraping API converting any URL to LLM-ready markdown(0.95 · 2 sources)
- Google Gemini 2.5 Flash·publicly released on 2025-04-09 by Google DeepMind — Gemini 2.5 Flash with controllable thinking budget + 1M context(0.95 · 2 sources)
- Groq LPU·publicly released on 2024-02-19 by Groq — language processing unit inference(0.95 · 2 sources)
- Mistral Magistral·publicly released on 2025-06-10 by Mistral AI — first Mistral reasoning model (Magistral Small open-weight + Medium API)(0.95 · 2 sources)
- Mistral Medium 3·publicly released on 2025-05-07 by Mistral AI — Medium-tier proprietary model balancing cost + performance(0.95 · 2 sources)
- NVIDIA Llama Nemotron·publicly released on 2025-03-18 by NVIDIA — Llama Nemotron family fine-tuned for reasoning + agentic workflows(0.95 · 2 sources)
- Qwen 3·publicly released on 2025-04-29 by Alibaba — Qwen 3 family (0.6B-235B), hybrid thinking mode, 128k context(0.95 · 2 sources)
- Reka Core·publicly released on 2024-04-15 by Reka AI — frontier multimodal LLM (text + image + audio + video input)(0.95 · 2 sources)
- Replit Agent·publicly released on 2024-09-05 by Replit(0.95 · 2 sources)
- xAI Colossus·publicly released on 2024-09-02 by xAI — 100,000-GPU H100 supercomputer (Memphis, TN) for training Grok models(0.95 · 2 sources)
- Group Relative Policy Optimization (GRPO)·introduced in paper DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models (Shao et al., 2024)(0.92 · 3 sources)
- Kahneman-Tversky Optimization (KTO)·introduced in paper KTO: Model Alignment as Prospect Theoretic Optimization (Ethayarajh et al., 2024)(0.92 · 3 sources)
- LiveCodeBench·introduced in paper LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code (Jain et al., 2024)(0.92 · 3 sources)
- MMLU-Pro benchmark·introduced in paper MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmark (Wang et al., 2024)(0.92 · 3 sources)
- Odds Ratio Preference Optimization (ORPO)·introduced in paper ORPO: Monolithic Preference Optimization without Reference Model (Hong et al., 2024)(0.92 · 3 sources)
- Simple Preference Optimization (SimPO)·introduced in paper SimPO: Simple Preference Optimization with a Reference-Free Reward (Meng et al., 2024)(0.92 · 3 sources)
- Anthropic Claude Memory·publicly released on 2025-09 by Anthropic — Claude memory referencing past chats (Team Sep 11 2025; Enterprise Sep 18 2025; Pro & Max Oct 23 2025; all plans Mar 2 2026)(0.90 · 2 sources)
- Anthropic Claude Sonnet 4.5·publicly released on 2025-09-29 by Anthropic — production model improvements over Sonnet 4 with extended-thinking + better coding(0.90 · 2 sources)
- Anthropic Skills·publicly released on 2025-10-16 by Anthropic — Claude can use named skills (curated capability packs)(0.90 · 2 sources)
- ChatGPT Atlas·publicly released on 2025-10-21 by OpenAI — AI-native web browser with built-in ChatGPT agent(0.90 · 2 sources)
- Google Imagen 4·publicly released on 2025-05-20 by Google DeepMind — Imagen 4 text-to-image with significantly improved text rendering(0.90 · 2 sources)
- OpenAI Sora 2·publicly released on 2025-09-30 by OpenAI — Sora 2 text-to-video with synchronized audio + improved physics fidelity(0.90 · 2 sources)
- xAI Grok 4·publicly released on 2025-07-09 by xAI — Grok 4 frontier model + Grok 4 Heavy multi-agent variant(0.90 · 2 sources)
- Anthropic Claude Haiku 4.5·publicly released on 2025-10-16 by Anthropic — fast tier Claude 4.5 family (text + vision + extended thinking)(0.85 · 2 sources)
- Anthropic Files API·publicly released on 2025-03-25 by Anthropic — file upload + reference API for Claude, supports PDF + image + spreadsheet(0.85 · 2 sources)
- ByteDance Doubao 1.5 Pro·publicly released on 2025-01-22 by ByteDance — MoE LLM matching GPT-4o quality on Chinese benchmarks, lower inference cost(0.85 · 2 sources)