Pinned Tweet
Open-weight models are ready for serious coding. But juggling providers, API keys, tools, and unpredictable costs still makes adoption harder than it should be.
We’re opening early access to Otari Subscriptions, to make open-weight models the easy choice for coding.
Try Otari for free at otari.ai or self-host it from github.com/mozilla-ai/otari.
Inkling 2 Small by @thinkymachines is now available on Otari.
With 594B parameters (17B active), it’s currently the 4th highest-scoring open model on Artificial Analysis Intelligence Index, with a score of 31.
It’s also the highest-scoring open model developed outside China.
1
3
201
Mozilla AI retweeted
We also opened Otari Subscriptions for early access: otari.ai/subscriptions
Another busy week for Otari:
- Web search and code execution out of Beta, plus @perplexity_ai as a search backend
- Per-end-user budgets and usage on service keys, and more accurate cost tracking
- Model listings & context lengths read from your @vllm_project & SGLang servers
We also opened Otari Subscriptions for early access: otari.ai/subscriptions
Mozilla AI retweeted
Just submitted ModelSwitch Lab to Mozilla's #DevsWildingOnline Hackathon
Choosing the right AI model shouldn't be guesswork.
I built ModelSwitch Lab to help developers test open-weight models, compare accuracy, speed, reliability and cost, and switch models based on real evidence.
Powered by @mozilla Otari, I implemented:
✅️Qwen & Llama model evaluation
✅️Real automatic model fallback
✅️MCP integration
✅️Native web fetch & code execution
✅️Gateway budget enforcement
✅️Request tracing & performance comparisons
6 advanced Otari capabilities verified with real execution evidence.
modelswitch-lab.vercel.app
Built, tested and submitted. Excited to see what comes next!
Today 15h CET, we'll be running our first office hour with the great @Marcus_Rein_.
If you have questions about Otari or what we do at Mozilla AI join us on Discord.
No plans, just chatting, testing, coding, and answering questions.
Join our Discord: discord.com/invite/QaA5NTEVW
2
94
Spots are limited. Request early access: otari.ai/subscriptions
1
1
66
More on why we're launching it: blog.mozilla.ai/otari-subscr…
2
64
Mozilla AI retweeted
open-weight models caught up. the barriers to adoption are what kept developers on the closed one.
Otari Subscriptions from @MozillaAI fixes that: one plan, one API key, any harness that takes a custom OpenAI or Anthropic endpoint. early access, live now. otari.ai/subscriptions
Meet transcribe.cpp 🎙️
A new open-source C/C++ speech-to-text inference library with:
⚡ GPU acceleration (Metal, Vulkan & CUDA)
🧩 Support for multiple GGUF STT models
💻 Fast, portable local transcription
Think "llama.cpp for speech".
👀 Read the article: blog.mozilla.ai/announcing-t…
Product work shouldn't mean juggling dozens of tabs.
See how we use Octonous at @MozillaAI to turn Slack feedback into GitHub issues, summarize PRs, and search across connected tools, all while staying in control.
👀 Read the full article: blog.mozilla.ai/using-octono…
Image classification is now available in encoderfile 📸
Deploy image classification models with the same zero-dependency approach used for text encoders.
This release includes:
- Image classification via CLI, HTTP/S, and gRPC
- Flexible preprocessing with Lua scripts
Object detection and image segmentation are next.
👀 Read the full post and let us know what you'd like to see next: blog.mozilla.ai/image-classi…
#OpenSource #AI #ComputerVision
7
297
What is an LLM Control Plane, and why does it matter for modern AI teams?
In our latest blog post, Anushri Gupta (Founder, DevRel at Mozilla.ai) breaks down the challenges of running LLMs in production and explains how a control plane goes beyond gateways to deliver the observability, governance, and reliability that today’s AI infrastructure demands.
Whether you’re building with LLMs now or planning for the future, this piece is a must-read.
Read the full article: blog.mozilla.ai/what-is-an-l…
#AI #LLM #MLOps #AIInfrastructure #MozillaAI #otari-ai
Can healthcare organizations build better AI without sharing sensitive patient data?
On June 10 at SIIM26 (@SIIM_Tweets), @johnpdickerson, CEO of @MozillaAI, will share how institutions can collaborate on AI models while keeping patient and provider data local and secure.
📍 Pittsburgh, PA
🗓️ June 10
See the session: link.mozilla.ai/siim26
2
141
Less than one week to go.
If your team spends too much time switching between tools, chasing updates, and repeating the same tasks, this webinar is for you.
Join Caroline Bohu from @MozillaAI for a live look at how AI agents can help.
Save your spot: link.mozilla.ai/octonous-jun…
5
116











