Lightnews — Scholar-powered news

gpt-oss is not for developers. It’s for agents. | Tigris Object Storage

Tigris Data @tigrisdata.com · 10d

Want to build a private agent but don't know which model you should pick? Try gpt-oss! We have a deep dive into the gorey details on our blog: www.tigrisdata.com/blog/gpt-oss/

Discover why OpenAI's gpt-oss model family is ideal for building reliable and safe AI agents, not just for developers.

Flux Kontext vs Nano Banana | Tigris Object Storage

Tigris Data @tigrisdata.com · 11d

Read the full article 📖 www.tigrisdata.com/blog/flux-ko...

We compared Flux Kontext, Nano Banana and Image GPT models. Find out which one is the best for your use case.

Tigris Data @tigrisdata.com · 11d

We compared Flux Kontext, Nano Banana, and GPT-Image-1 for image gen. Result:

◆ Flux is the best for most use cases and also the fastest & cheapest.
◆ Nano Banana is very close behind.
◆ GPT-Image-1 is the best at detail.

Compare the models yourself 👉 www.image-models.dev

Announcing the Tigris Storage SDK | Tigris Object Storage

Tigris Data @tigrisdata.com · 17d

If you want to read more about our philosophy and see side-by-side examples with the AWS SDK, check out our blog post about it: www.tigrisdata.com/blog/storage...

Introducing the Tigris Storage SDK for JavaScript and TypeScript - a simplified alternative to AWS S3 SDK with environment-based configuration, error-as-values pattern, and streamlined object storage…

Tigris Data @tigrisdata.com · 17d

If you want to get started, check it out on NPM: www.npmjs.com/package/@tig...

Tigris Data @tigrisdata.com · 17d

When we dreamed this SDK into existence, we wanted to make sure that your code is easy to read, easy to understand, and easy to work with. We took all of the most common uses of object storage and reduced them down to their smallest possible form so that you just focus on what really matters:

We do our billing with Prometheus | Tigris Object Storage

Tigris Data @tigrisdata.com · 17d

What if using object storage was easy as calling "get", "put", "list", and "delete"? Imagine what could be if simple operations were just simple.

You don't have to imagine it anymore, today we're announcing the Tigris Storage SDK. More in the 🧵

2 5

Tigris Data @tigrisdata.com · 19d

Read more on the blog: www.tigrisdata.com/blog/billing...

How Tigris uses Prometheus metrics for usage-based billing instead of traditional event-driven systems, reducing complexity while ensuring transparency and customer-friendly error handling.

Tigris Data @tigrisdata.com · 19d

At Tigris, we use our metrics system to do usage-based billing. It's much simpler than it sounds. This lets us use data we already collect, show to you, and rely on implicitly to keep Tigris functioning in our billing process. This makes it easier for us to make Tigris shine!

Unlimited Message Retention with BufStream and Tigris | Tigris Object Storage

Tigris Data @tigrisdata.com · 25d

$0.22? Our quarterly numbers may never fully recover from the shock of such a boon!

Really though if you run into problems, ping me! I'll be happy to chime in :)

1 2

Tigris Data @tigrisdata.com · 25d

🫡

2

Tigris Data @tigrisdata.com · 25d

Kafka alternatives are evolving fast! 🚀

Bufstream brings a fresh take on event streaming — and with Tigris object storage powering message retention, it’s built for scale + durability.

Learn more: www.tigrisdata.com/blog/bufstre...

Get unlimited message retention and truly global operation with BufStream and Tigris.

Changelog | Tigris Object Storage Documentation

Tigris Data @tigrisdata.com · Sep 10

🔗 Follow what we're building: www.tigrisdata.com/docs/changel...

Tigris Data @tigrisdata.com · Sep 10

Check out our new changelog, a balance of feature announcements, blog posts, and API changes.

Distributed Training with LanceDB and Tigris | Tigris Object Storage

Tigris Data @tigrisdata.com · Sep 9

Read more on the blog! www.tigrisdata.com/blog/lancedb...

Discover how to efficiently train large-scale ML models using LanceDB and Tigris. This guide details streaming massive datasets directly from object storage to PyTorch, simplifying distributed…

Distributed Training with LanceDB and Tigris | Tigris Object Storage

Tigris Data @tigrisdata.com · Sep 9

There's two patterns for storing data in Tigris: map style and iterable style. Map style is great for finite datasets, but sometimes the data is so big that you can't even fit an index of that data into memory. That's when you want the iterable style.

Discover how to efficiently train large-scale ML models using LanceDB and Tigris. This guide details streaming massive datasets directly from object storage to PyTorch, simplifying distributed…

Distributed Training with LanceDB and Tigris | Tigris Object Storage

Tigris Data @tigrisdata.com · Sep 9

Tigris is a great place to store your AI training data. When you use Lance on top of Tigris, you can do your training directly from object storage without having to copy it to your servers first. We made a deep dive on how you can do this and what tools you have available.

Discover how to efficiently train large-scale ML models using LanceDB and Tigris. This guide details streaming massive datasets directly from object storage to PyTorch, simplifying distributed…

How LogSeam Searches 500 Million Logs per second | Tigris Object Storage

Tigris Data @tigrisdata.com · Sep 4

🔗 Read more about LogSeam's 500M logs/sec queries on the blog: www.tigrisdata.com/blog/case-st...

LogSeam transforms log management with Tigris, achieving massive cost savings and performance improvements in their security data lake architecture.

Global Replication Made Easy | Tigris Object Storage

Tigris Data @tigrisdata.com · Sep 4

We were shocked watching LogSeam query 10TB of logs in seconds.

How?

- Refactored to Parquet & compressed 100x

- Data moved close to compute w/ @TigrisData

- 100s of nodes spin up per query

👉 500M+ logs/sec, 1.5TB+/sec search
💸 40–80% lower cost

🔗 link to the blog in reply

1 2

Tigris Data @tigrisdata.com · Sep 3

Read more on our blog: www.tigrisdata.com/blog/talks/2...

Supplemental materials for the DevOpsDays Halifax 2025 talk and some extra info that didn't make the cut.