Long-form reasoning
Wren works through layered problems that go well past pattern-matching or text completion.
Meet Wren
Wren is reasoning you can hand real work to — whether you're thinking alone or shipping with a team of thousands.
What Wren is good at
Visual analysis
Read and describe nearly any still image — scanned notes, dense charts, product photography.
Code generation
Scaffold interfaces, turn screenshots into structured records, or trace a bug through a large repo.
Multilingual work
Translate live between dozens of languages, drill grammar, or draft copy for several markets at once.
New
Introducing Wren 4 Meadow
A new intelligence ceiling at the latency and price high-volume teams actually budget for.
Read the release noteThe Wren model family
Right-sized for the task in front of you — three models tuned to different points on the speed-and-depth curve.
Intelligence
Deepest
Summit
Our highest-performing model — multi-step analysis, long research threads, and higher-order math and code.
Balanced
Meadow
The best mix of judgment and throughput for the work that runs every day.
Light & fast
Ember
Our quickest model, built for lightweight actions at industry-leading latency.
Cost
Why teams pick Wren
Alderline builds frontier models the slow way, and publishes what we find along the way.
Secure
Deployable through two major clouds, with SOC 2 Type II certification and HIPAA-eligible configurations for regulated workloads.
Featured paper
Many-turn prompt saturation
Featured post
What breaks first when you red-team a reasoning model
Trustworthy
Wren pairs best-in-class jailbreak resistance with misuse controls, so the brand risk stays where it belongs — nowhere.
Featured paper
Adversarial probing of language models: methods, scaling behavior, and what we learned
Featured paper
Measuring and reducing disparate outcomes in model decisions
Reliable
Teams route business-critical work through Wren because hallucination rates stay low and accuracy holds across very long documents.
Featured post
Wren's disposition
Featured paper
Question decomposition improves the faithfulness of generated reasoning
Teams building on Wren
solane
Verdant
NVL
quillon
Bramblewick
Cobalt Bay
Larkfield
Meridian
Rail Tessera Umbra Foundry
Rail Tessera Umbra Foundry
Talk to Wren
Wren is quick, capable, and genuinely conversational. Bring it the messy first draft and work forward from there.
Try WrenBuild with Wren
Wire the API into your product and your customer workflows, and let the model carry the parts nobody wants to hand-code.
Get API keysTake Wren with you
Ask Wren anywhere you happen to be. Sketch an idea, settle a question, or point the camera at a chart and get it read back to you — frontier reasoning in a coat pocket.
Company news
See allAnnouncements
Widening our model safety bug bounty
Aug 8, 2025
Announcements
Wren is now available across Aotearoa
Aug 2, 2025