mts @reflection_ai working on mirror
nyc
Joined December 2022
- Tweets434
- Following1.5K
- Followers515
- Likes17.3K
Simon retweeted
Mirror is a Rust agent core with a python TUI using an RPC. The same agent core can run headless behind an HTTP API just as easily as in your terminal.
Every session is a single append only log, and everything else (the transcript, what the model actually sees, token usage, extension state) is just a fold over that log.
Compaction appends a new context window with a summary and the recent turns, and a rewind is a fork that points back at the parent’s log up to where you rewound, so nothing ever gets copied or thrown away.
Simon retweeted
Ok can finally talk about what I’ve been working on at Reflection. Introducing Mirror, our own open source coding agent optimized for Beam.
I started the mirror project when I first joined Reflection, and it has gone through many iterations since then.
Install it today with: uv tool install reflection-mirror
github.com/reflection-oss/mi…
Simon retweeted
Our team is addicted to creating Extensions. Some or silly like catjam memes (looking at you @bhutanisanyam1 ) and some are productive like @depthfirstsearc feature below. Some are aesthetic like a world clock that I added to my Mirror. Go build your own and give us your feedback!
developers.reflection.ai/mir…
Simon retweeted
Reflection’s first model, Beam, scores 30 on the Artificial Analysis Intelligence Index, placing it #2 among US open weights models
Beam is an open weights model with 501B total and 23B active parameters, released on @huggingface
Congratulations to @reflection_ai on the release!
Key takeaways:
➤ Second among US open weights models: Beam scores 30, behind only Meta's Muse Spark Open (39) and ahead of yesterday’s NVIDIA Nemotron 3.5 Super VL (26) release. Outside China, only Muse Spark Open and the UAE's K2 Horizon (31) score higher
➤ Token efficient for its intelligence: Beam uses ~26k output tokens per Intelligence Index task compared to K2 Horizon, one point ahead, which uses twice as many (~52k). Proprietary models still set the frontier, with GPT-6 Astra (max) scoring 53 on the same budget (~27k)
➤ Strong on agentic workflows, weaker in the terminal: Beam scores 48.5% on AutomationBench-AA, 11 percentage points ahead of K2 Horizon (37.2%) and level with Qwen3.8 27B (48.2%). Terminal-Bench 4.0 is a weak point, where Beam scores 3.0%, behind Qwen3.8 27B (5.6%) and Muse Spark Open (8.1%)
➤ Knows more, guesses more: On AA-Omniscience it answers 33% of questions correctly, well ahead of K2 Horizon (18%) and Qwen3.8 27B (16%). But its hallucination rate is 47%, against 26% for K2 Horizon and 30% for Qwen3.8 27B, giving Beam a net index score of 1.5, ahead of both peers (−3.0 and −10.0)
Other model details
Context window: 1M tokens
Modalities: Text
Parameters: 501B / 23B
Simon retweeted
Beam is now available across platforms, with open weights under the Apache 2.0 license - in collaboration with an ecosystem of 40+ partners.
We're releasing:
- Weights, tokenizer, and reference implementation
- BF16, FP8 and NVFP4 weights and a DFlash2 drafter
- The open source harness we built and trained Beam on: Mirror CLI
Weights and model card on Hugging Face: huggingface.co/reflection
Simon retweeted
Today, we’re introducing Beam, Reflection’s first model.
A year ago, we set out to build a frontier open model before we had much of the team or infrastructure required to do it.
Since then, Reflection has grown 10x. We assembled an extraordinary technical team of researchers and engineers I feel privileged to call my colleagues, alongside a world-class group across recruiting, business, policy, operations, and the rest of the company whose work made all of this possible.
At the same time, we built the science, data, software, and infrastructure behind the model itself. In July, we started pretraining Beam and weeks later, we had a 500B-parameter model at the frontier of Western open intelligence—particularly strong at agentic coding, with remarkable reasoning efficiency.
What makes me proudest, though, is our team. The relentless focus, speed of execution and the willingness to own and solve whatever problem was in front of us. And, most importantly, the commitment of this team to our mission.
Beam is our first milestone toward our mission of building open intelligence and making it accessible to everyone.
We’ll release Beam’s weights, model card, and tech report later this month.
And we’re already training the next, larger model.
Over the past year, we built the team and the engine to propel us to the frontier and that’s what makes me so excited about what comes next.
Introducing Beam: a highly efficient agentic open model with 501B total parameters and 23B active.
- Frontier reasoning efficiency
- Advances the Western open frontier on coding & agentic tasks
- Trained end-to-end from scratch
Full weights release this month.
Learn more about Beam: reflection.ai/beam
Simon retweeted
I am so proud of Beam.
Beam is Reflection's first open-weight model. It's the result of a ~year of hard work, ingenuity, and camaraderie of one of the most incredible teams I've ever seen in AI 🔥
I joined Reflection last November. Pretraining in particular was so nascent – only 5-10 people and a bit of code! We didn't yet have a crawler, or a resilient GPU infra, or a deduplication pipeline, or in-house evals, or even sufficiently big MoEs. All that started in earnest in 2026, in one of the most satisfying sprints I've ever had the pleasure to live through.
What we nailed, I think, was a perfect blend of scientific rigor, startup intensity, team iteration, and ambition. We did not YOLO decisions in pretraining, that tends to blow up in your face when you scale up 🙂 But we also made sure to move fast and play to our strengths as a nimble, high-ownership, high-trust team. The resulting system of systematic iterations and fast cross-team back-and-forths really paid off.
The midtraining, RL, and post-training teams really really cooked here on top of the pretrained base. The announcement below details the sheer scale and payoff of that investment, I'll let the team cover it, it's remarkable. On many agentic coding capabilities the model kept going up without any sign of a plateau.
Feel free to read more in the announcement today – but later this month we will also drop the full weights release under Apache 2.0, a detailed tech report with the cool LLM science, and partner integrations with the OSS ecosystem. It's an open model after all. I strongly believe open intelligence is the future of AI, this is why I'm here.
And I encourage y'all to join us on the open side 😉
Introducing Beam: a highly efficient agentic open model with 501B total parameters and 23B active.
- Frontier reasoning efficiency
- Advances the Western open frontier on coding & agentic tasks
- Trained end-to-end from scratch
Full weights release this month.
Learn more about Beam: reflection.ai/beam
Simon retweeted
Introducing Beam: a highly efficient agentic open model with 501B total parameters and 23B active.
- Frontier reasoning efficiency
- Advances the Western open frontier on coding & agentic tasks
- Trained end-to-end from scratch
Full weights release this month.
Learn more about Beam: reflection.ai/beam
Simon retweeted
Certus AI (@certus_ai) is a voice agent for restaurants that answers every call, takes orders, deliveries, and books reservations — integrated with platforms like Toast, Square, UberEats & DoorDash.
ycombinator.com/launches/O1Q…
Congrats on the launch @Gurveersingh299, @AdamGamieldien & @_apisaac!
Simon retweeted
Today we’re excited to introduce Bland TTS, the first voice AI to cross the uncanny valley.
Several months ago, our team solved one-shot style transfer of human speech. That means, from a single, brief MP3, you can clone any voice or remix another clone’s style (tone, cadence, or pronunciations, etc.).
With Bland TTS:
• Creatives can turn text into realistic sound effects or AI voice tracks with precise control over style and emotion.
• Developers can plug in our TTS API to build voice apps fully customized to their liking.
• Enterprises can launch AI-driven support lines that sound so natural — customers will save you as a contact (try talking to it on our site).
🧵Check out the thread below or visit our website to get started for free
Simon retweeted
What’s actually happening under the hood of voice AI?
We just dropped 3 deep dives on LLMs:
How they think (tokens → transformers)
How we keep them secure
How we stop jailbreaks
If you read this stuff for fun, we’re hiring AI Deployment Engineers.
$10K for a referral. 💸
Links in comments. 👇
Simon retweeted
Hiring a ton of forward deployed AI engineers right now.
The core thing we're looking for:
- People who want to own projects from start to finish (organized, PM skills)
- Are excited to prompt engineer and push LLMs to their limits
- Are technical (know how APIs integrate and work)
Former / future founders, people with weird backgrounds, and anyone else who is interested, please reach out!
Simon retweeted
🚨 SMS is here!
You can now build AI agents that text, one of our most requested features from enterprise teams.
Every enterprise Bland number (yours or ours) can now send + receive SMS like a real contact.
No extra setup.
Here’s how it works:
• Pick a number
• Connect to a prompt or pathway
• You’re live — $0.015 per message
🔒 Plus: we’re now fully GDPR compliant, on top of SOC 2 Type II + HIPAA.
More updates coming soon. 🚀
Simon retweeted
🏈 It didn’t cost $8M. And you never even saw our “Super Bowl ad”. But everyone's convinced they did... and it went viral anyways.
Over the past week, universities have been showing our ad to students.
Americans from across the country have been calling in to talk to our AI.
And enterprises flooded @Bland’s website.
💰But how did a startup afford an $8,000,000 ad slot?
We didn’t. And you probably never even saw it live 🧵
Simon retweeted
Love is complicated. Understanding customers over the phone shouldn’t be.
Meet Babel, our new real-time multilingual transcription engine, built to understand conversations across languages—because nothing should get lost in translation.
Watch the demo, and happy (early) Valentine’s Day.
Simon retweeted
We just open-sourced the best voicemail detection model in the world.
98.5% accuracy. No awkward messages. No wasted calls.
Try it & read the full breakdown: shorturl.at/grl9q
