2025: A Year of Breakthroughs
From 9M tokens in 2023 to 1.5B in 2025 — we've grown 166x together. Here's what we built.
1.5B+
Tokens Processed
50+
AI Models
6,000+
App Integrations
1,750
T/s LeemerLite
Thank You for More Than a Year of LeemerChats
Thanks for sticking with two proudly self-declared novices. Over 24+ months you burned 1.5 billion tokens with us—coaching ideas into products, debugging real life, and proving that community beats polish. It's been absolutely crazy, and we're going even crazier.
When we opened this version of LeemerChat we promised to stay learners first. We called ourselves novices on purpose so we could keep tinkering without ego. That experiment worked because you showed up with curiosity and patience every single night. Now we're scaling infrastructure that handles 1B tokens per day—and we're just warming up.
Together We've Processed
Tokens
Every question, every idea, every late-night debugging session. You made this possible. And we're just getting started.
From 9M in 2023 → 40M in 2024 → 500M in H1 2025 → 1B in H2 2025
We're scaling like crazy. Ready for 1B tokens per day.
New Features We Shipped Together
From Auto Research to Deep Research 80B, we've revolutionized how you explore information. Our research agents work in the background, delivering comprehensive reports to your inbox.
Auto Research
Background research agents that deliver results to your inbox
Leemer Deep Research 80B
Multi-hop reasoning with 256K context for complex queries
Email Research
AI-powered research delivered directly to your email
Firecrawl Web Search
Real-time web access with intelligent citations on ALL models
Email Agent
AI copilot in your inbox at agent@leemerchat.com
PowerCode
AI coding agent with GitHub integration
LeemerLabs Foundry
Ireland's first custom LLM creation studio
LeemerGLM-106B
24-expert Mixture-of-Experts model
LeemerLite
1,750 T/s sandbox, no login required
Agent-Leemer-K2
Zapier MCP with 6,000+ app integrations
Durable Generation
Survives page refreshes & browser closures
Voice Mode
Real-time voice assistant with WebRTC
Writer & Docs
Harvard citations, versioning, and folders
Auto-Research Podcasts
Generate audio discussions from research
Research Spotlight
Try Auto Research and Deep Research — our most powerful features that deliver comprehensive reports while you focus on other tasks.
Try LeemerLite: 1,750 Tokens/Second
Need instant answers with zero friction? LeemerLite is our blazing-fast sandbox powered by Groq's LPU Inference Engine — running at 1,750 tokens per second.
Lightning Fast
1,750 T/s with Groq's tensor streaming
No Signup
Jump in instantly, no account needed
Local History
14-day client-side storage, privacy first
💡 Perfect for quick questions during calls, rapid prototyping, or when you need answers right now. Keep it pinned alongside your main workspace for instant access to world-class AI without the weight.
The Journey: V3 → V4
Consider this our goodbye letter to the V3 era. It's a scrapbook of what we learned together and a promise that the experiments continue.
2023: Built as a Backup
Repath assembled a safety net when ChatGPT outages made work grind to a halt. Early friends used it to keep publishing overnight. 9M tokens sparked the journey.
2024: Renamed LeemerChat
The scrappy backup turned into a unified workshop with writer, research, podcast, and sharing flows. 40M tokens by end of year hardened the pipeline.
H1 2025: The Acceleration
500M tokens in the first half alone. Multi-model orchestration, email agents, and foundry launch proved the architecture could scale. The floodgates opened.
H2 2025: The Billion Era
1B tokens in six months. PowerCode, LeemerLite at 1,750 T/s, Durable Generation, and Agent-Leemer-K2 with 6,000+ integrations. We're ready for 1B tokens per day.
V4.5: Research & Email Automation
Auto Research, Email Research, and Leemer Deep Research 80B launched. Background agents keep working after you close the tab—results arrive in your inbox.
V4.7: Firecrawl Web Search
Real-time web search with intelligent query optimization, enhanced citations, and persistent source tracking across all models.
V4.8: The Smoother Experience
IKEA-inspired UI, frosted glass interfaces, durable generation that survives page refreshes, and LeemerGLM-106B-A22B launch.
V4.9: PowerCode & Agent-Leemer-K2
AI-powered coding agent with GitHub integration, LeemerLite sandbox at 1,750 T/s, Agent-Leemer-K2 connecting 6,000+ apps via Zapier MCP, and now surpassing 1.5B tokens processed together!
Welcome to the New Lineup
This year we welcomed incredible new models to LeemerChat. From GPT-5.1 to Claude 4.5 Sonnet, Qwen, and beyond — you now have access to the most powerful AI models on the planet.
GPT-5.1 Chat
OpenAI
The most capable OpenAI model with enhanced reasoning and 1M token context
Claude 4.5 Sonnet
Anthropic
Anthropic's latest with superior writing, analysis, and deep refactoring
Gemini 3 Pro
State-of-the-art benchmarks: 37.5% on Humanity's Last Exam, 1M-token context, full multimodal
Qwen3 235B A22B
Alibaba
Massive 235B MoE with vision reasoning—powers Leemer Deep Research
DeepSeek V3.2 Speciale
DeepSeek
Open-source powerhouse with chain-of-thought reasoning and coding excellence
Grok 4 Fast
xAI
Lightning-fast responses with real-time knowledge from xAI
Shoutout to the Legends Making This Possible
A massive thank you to the open source AI labs, model providers, and infrastructure partners who power LeemerChat's speed and intelligence. We wouldn't exist without you.
Special Thanks: OpenRouter
OpenRouter is the unified API that powers our multi-model orchestration. They give us access to 200+ models from every major provider with one consistent interface. Every model switch, every failover, every speed optimization—OpenRouter makes it seamless. Thank you for being the backbone of LeemerChat's model marketplace.
#1 Most Used
Qwen
Qwen3 VL 30B A3B
Vision + 131K context — powers multimodal workflows
Qwen3 32B
Our default model — fast, capable, reliable
Qwen3 Next 80B A3B
Powers Leemer Deep Research with 131K context
Alibaba's Qwen family runs the show. From vision to reasoning, these models define LeemerChat's core intelligence.
#2 Runner Up
Moonshot
Kimi K2
251K context — beats GPT-5.1 in many benchmarks
Kimi K2 Thinking
World's best open-source reasoning model
Kimi Linear 48B
1M token context — for massive documents
Moonshot AI's Kimi family excels in Chinese and long-context work. Essential for Leemer Heavy Fast synthesis.
#3 Bronze
DeepSeek
DeepSeek V3
Exceptional reasoning at a fraction of the cost
DeepSeek V3.2 Speciale
131K context — chain-of-thought excellence
DeepSeek R1
Reasoning powerhouse for complex problems
DeepSeek proves open source can compete with the best. Their models power many of our advanced reasoning flows.
Thank you to the open source community
Alibaba (Qwen), Moonshot AI (Kimi), DeepSeek, Meta (Llama), Mistral AI, Google (Gemma), Groq, xAI, and countless others pushing open source AI forward. Your work makes LeemerChat possible. We're honored to build on your foundations.
A Note from the Heart
Repath "Ray" Khan, Founder
Hey friends,
As I sit here in Waterford reflecting on this wild ride, I'm overwhelmed with gratitude—and a bit of disbelief. When we built LeemerChat in 2023 as a backup when ChatGPT outages made work grind to a halt, we never imagined we'd hit 1.5 billion tokens by the end of 2025. It's been absolutely crazy. And we're going crazier.
You believed in us before we had fancy features. You stuck with us through the bugs, the late-night maintenance windows, and the ambitious experiments that sometimes went sideways. From 9M tokens in early 2023 to 40M in 2024, then 500M in H1 2025, and another 1B in H2—every transcript, every bug report, every late-night chat kept the lights on.
I want you to know that LeemerChat exists because of you. The community you've built here is something we never expected but will forever cherish.
From the V3 era to now V4.9 with PowerCode, LeemerLite running at 1,750 T/s, Deep Research with multi-model orchestration, Email Agents delivering to your inbox, Agent-Leemer-K2 connecting 6,000+ apps via Zapier MCP, and Gemini 3 Pro breaking benchmark records—every breakthrough happened because you showed up, pushed us to be better, and never let us settle.
We're ready for 1B tokens per day. We're scaling infrastructure, refining agents, and we're open to funding that aligns with our mission: make advanced systems feel friendly to novices and unstoppable for the ambitious.
Happy Holidays to you and yours. Here's to another year of building the impossible together.
With love and gratitude,
Ray
Agent-Leemer-K2: Connect Everything
Agent-Leemer-K2 is our Model Context Protocol (MCP) integration that connects LeemerChat to 6,000+ apps via Zapier. Email, calendars, databases, CRMs, social media—if Zapier supports it, Agent-Leemer-K2 can orchestrate it.
Multi-App Actions
Chain actions across Gmail, Slack, Notion, and 5,997 more apps
Real-Time Triggers
React to webhooks, schedules, and external events instantly
No-Code Workflows
Describe what you want in plain English, let K2 build it
Example: "When I get an email from a client, summarize it, create a Notion task, update my CRM, and post a Slack notification." Agent-Leemer-K2 handles the rest.
Get Ready for V5
Autonomous agents that never sleep. Multi-agentic systems that collaborate in real-time. Better connection options. Native mobile apps. Voice-first workflows. We're building the future of AI workspaces—and it's going to be wild.
Autonomous Agents
Long-running tasks that execute while you sleep
Multi-Agent Collaboration
Specialist agents working together in parallel
Native Mobile Apps
Full-featured iOS and Android experiences
Enhanced Integrations
Deeper connections with your favorite tools
Same heart, sharper tools, still building in public.
Explore new models, try new features, and keep building amazing things. We're scaling to 1B tokens per day. V5 is coming. And we're open to funding.