Orchestra

Orchestra trains company-specific AI models that are 100x smaller than frontier from the work employees and agents already do.

[Orchestra @orchestradotai]

How fit is your AI? Take the quiz:

https://t.co/NM6RONC52r

[Luis Manrique @lluismanrique]

GLP-1 for your AI bill. On the streets of San Francisco today.

https://t.co/aSyVVMH9lQ

[Luis Manrique @lluismanrique]

GLP-1 for your AI bill. On the streets of San Francisco today.

https://t.co/aSyVVMH9lQ

[Orchestra @orchestradotai]

Orchestra (https://t.co/BITWcPCxS6) automates fine-tuning open-weight models on your proprietary company data.

Own your intelligence & outperform frontier models on specialized tasks, at much lower cost and latency.

[Luis Manrique @lluismanrique]

in our latest 'backstage' conversation at @orchestradotai , our founding engineer @ripertmat covers:

why companies need evals

why you can start with harness engineering and move to building training environments

how orchestra ties all this together

[Luis Manrique @lluismanrique]

in our latest 'backstage' conversation at @orchestradotai , our founding engineer @ripertmat covers:

why companies need evals

why you can start with harness engineering and move to building training environments

how orchestra ties all this together

[Orchestra @orchestradotai]

@GradyAndersen @lluismanrique not all! we're in the wild west

[Luis Manrique @lluismanrique]

better luck next time Haiku!

we're seeing GPT-6 Luna maintain it's price/performance lead

[Orchestra @orchestradotai]

@OpenAI 's Luna models look to be superior up to XHigh and Max reasoning levels, but we'll be sharing a full breakdown of performance from our private benchmarks to see where this model shines.

[Orchestra @orchestradotai]

Anthropic's Claude Haiku 5.5 is live for Orchestra users (https://t.co/BITWcPD5HE)

We'll be sharing benchmarks this afternoon; Anthropic deployed their new tokenizer to this model as well which historically has led to >30% increase in overall costs-per-task.

[Luis Manrique @lluismanrique]

https://t.co/DdkUO3pcZD

[Luis Manrique @lluismanrique]

We used Orchestra (@orchestradotai ) to evaluate the same GLM 5.3 Flash model across 9 different inference providers on our private benchmark.

Same model, same weights, same work.

But up to 6x difference in costs and 3x difference in performance.

[Luis Manrique @lluismanrique]

GLM 5.3 Flash has become a new workhorse for customers moving away from Anthropic

20x lower cost-per-task at the same or better performance as Sonnet 5.5

[Orchestra @orchestradotai]

Orchestra started September at YC Demo Day and ended the month welcoming employee number five.

You don’t need a PhD to sort the mail, and you shouldn’t pay for the biggest AI model for every task. We believe every company using AI should be accumulating intelligence, not accumulating API bills.

[Yahia Bakour @mynameisyahia]

The https://t.co/lqMcRkRhm1 free tier is back.

You can now scrape, search, and turn the web into clean context for your agents, for free.

500 credits / mo

Go build something great!

[Luis Manrique @lluismanrique]

Excited to announce that @aamirpoonawalla and I are launching @understudylabs as part of @ycombinator 's S26 batch.

We help you cut your company’s Anthropic bill by 80% with no loss of performance.

Understudy is an inference cloud. We capture your AI traces, train smaller models, and automatically deploy them (only when they outperform the expensive model you are currently paying for).

Instead of renting the same frontier model forever, your AI infrastructure gets cheaper and better every time it runs.