Introducing Holo4

Models API

Multimodal models for real-world use

Process text, images, and documents, with computer-use capabilities for any digital environment.

New

Flagship

Holo4 27B

Dense, 27B

Best accuracy on long, multi-step tasks across web, desktop, and mobile.

85.2%on OSWorld
Context
256K
Input / 1M$0.04 cached
$0.40
Output / 1M
$3.00

New

Fast

Holo4 35B-A3B

MoE, 35B total, 3B active

Close to 27B accuracy, cheaper and faster. Open weights under Apache 2.0.

80.8%on OSWorld
Context
256K
Input / 1M$0.03 cached
$0.30
Output / 1M
$2.00

Previous generation, still served

Holo3.1 35B-A3B
holo3-1-35b-a3b
Context64KInput / 1M$0.25Output / 1M$1.80
Holo3 122B-A10B
holo3-122b-a10b
Context64KInput / 1M$0.40Output / 1M$3.00

Rate-limited free access to Holo3.1 35B-A3B. No credit card required. Pay-as-you-go for every model, with higher rate limits.

After you pay, it can take up to 15 minutes for the system to process your payment and credit your balance.

Frequently asked questions

Start with Holo4. Holo4 35B-A3B is the default for production computer-use workloads: longer context, strong accuracy, and low cost and latency. For denser flagship-class reasoning on complex multi-step tasks, use Holo4 27B. Earlier Holo3 models remain available on the same Models API if you need them. Switching only requires a model ID change.

Multimodal models for real-world use

Process text, images, and documents, with computer-use capabilities for any digital environment. Rate-limited free tier, pay-as-you-go beyond.