> ## Documentation Index
> Fetch the complete documentation index at: https://ubicloud.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# January 2025

export const PrList = ({ids = []}) => <sub>
    {ids.map((id, index) => <>
        {index > 0 && ", "}
        <a href={`https://github.com/ubicloud/ubicloud/pull/${id}`} target="_blank">
          <u>
            {index === 0 ? "PR " : ""}#{id}
          </u>
        </a>
      </>)}
  </sub>;

## AI inference endpoints (new)

You can now use our OpenAI compatible API endpoints through AI Inference on Ubicloud console. 

### 7 open source LLM and embedding models

* Ds-r1-qwen-32b (Text-generation) 
* llama-3-1-8b-it
* Llama-3-2-3b-it
* Llama-3-3-70b-it
* qwen-2-5-14b-it
* qwq-32b-preview
* E5-mistral-7b-it (Embedding)

Summary cards display each model's key properties in the console. 

<img src="https://mintcdn.com/ubicloud-00f14c0d/dkQS7vFTPxPT-89t/changelog/2025-01-ai-inference-endpoints.png?fit=max&auto=format&n=dkQS7vFTPxPT-89t&q=85&s=8e24e4d60af19c7f4c49f33e15c770f6" alt="Ubicloud AI inference" width="2764" height="2270" data-path="changelog/2025-01-ai-inference-endpoints.png" />

<Note>Models are priced between \$0.10 - \$0.60 per million tokens</Note>
<Tip>Ubicloud API endpoints cost \~10x less when compared to OpenAI models at \$4.40 - \$10.00. While different models, the open source models provide notable performance for their price.</Tip>

<PrList ids={[2640, 2554]} />

### Playground for testing models

<img src="https://mintcdn.com/ubicloud-00f14c0d/dkQS7vFTPxPT-89t/changelog/2025-01-ai-playground.png?fit=max&auto=format&n=dkQS7vFTPxPT-89t&q=85&s=6a74841554001ff349a17ca5957a7119" alt="Ubicloud AI playground" width="2760" height="1474" data-path="changelog/2025-01-ai-playground.png" />

## Managed PostgreSQL

### Reduced prices by 15-20%, as new hardware continues to improve price-performance\*\*

<img src="https://mintcdn.com/ubicloud-00f14c0d/dkQS7vFTPxPT-89t/changelog/2025-01-best-postgres-cloud-pricing.png?fit=max&auto=format&n=dkQS7vFTPxPT-89t&q=85&s=77da50c04b02b4f12486ed5c6aa40ad7" alt="Ubicloud Managed PostgreSQL pricing" width="1688" height="420" data-path="changelog/2025-01-best-postgres-cloud-pricing.png" />

<Note>Ubicloud standard-2 PostgreSQL instances now cost below \$50 per month.</Note>

<Tip>This is 4x lower than AWS RDS, while also delivering better performance, driven by new hardware and NVMe disks.</Tip>

### Minimum disk size reduced from 128 GB to 64 GB

<img src="https://mintcdn.com/ubicloud-00f14c0d/dkQS7vFTPxPT-89t/changelog/2025-01-pg-disk-size.png?fit=max&auto=format&n=dkQS7vFTPxPT-89t&q=85&s=e0333267c61091c30658a81f25aaadc8" alt="PostgreSQL disk sizes" width="2798" height="1712" data-path="changelog/2025-01-pg-disk-size.png" />

### Larger storage options introduced also in the US

<Note>For large storage options in PostgreSQL that exceed 16 TB, please contact us via support.</Note>

### Download PostgreSQL CA certs from Ubicloud console

This makes it easier to verify certs via SSL for more secure connections.

### ParadeDB new images: v0.14.1

ParadeDB got a major upgrade, now running on PostgreSQL block storage. Ubicloud managed ParadeDB has the updated images. Read more on the [ParadeDB blog.](https://www.paradedb.com/blog/block_storage_part_one)

[https://docs.paradedb.com/changelog/0.14.0](https://docs.paradedb.com/changelog/0.14.0)

<PrList ids={[2649, 2643, 2572, 2523, 2506, 2464]} />

## Sign in with Google and Github

**You can now use your existing Google or Github accounts to login to Ubicloud.** Passwordless login is also enabled.

<img width="300" alt="Social Login" src="https://mintcdn.com/ubicloud-00f14c0d/dkQS7vFTPxPT-89t/changelog/2025-01-social-login.png?fit=max&auto=format&n=dkQS7vFTPxPT-89t&q=85&s=0d24e8678bb4912ce95d9c8e2dc7fd0e" data-path="changelog/2025-01-social-login.png" />

<PrList ids={[2463]} />

## Simplified fine-grained access controls

Ubicloud comes with granular access controls that can be assigned to each user, for different resources. These Attribute Based Access Controls (ABAC) can now be managed much more simply through the console.

### Attribute Based Access Controls (ABAC) managed via console

<img src="https://mintcdn.com/ubicloud-00f14c0d/dkQS7vFTPxPT-89t/changelog/2025-01-auth-rbac.png?fit=max&auto=format&n=dkQS7vFTPxPT-89t&q=85&s=bff3af75dddb68c9cb21a5d917d48ced" alt="" width="2782" height="1344" data-path="changelog/2025-01-auth-rbac.png" />

### Step-by-step guide to user permissions & API access management

With major changes to how authorization is managed, we created a detailed documentation page, with example scenarios: [https://www.ubicloud.com/docs/security/authorization](https://www.ubicloud.com/docs/security/authorization)

### Granular access extended to APIs via personal access tokens

<img src="https://mintcdn.com/ubicloud-00f14c0d/dkQS7vFTPxPT-89t/changelog/2025-01-personal-access-tokens.png?fit=max&auto=format&n=dkQS7vFTPxPT-89t&q=85&s=360acd32d5ac4934a0e64db290244734" alt="" width="2810" height="1096" data-path="changelog/2025-01-personal-access-tokens.png" />

With these changes, we also deprecated our prior JSON based policy editor in the console.

<PrList ids={[2384, 2516]} />

## Managed Github Actions Runners

### Transparent Cache enabled by default for all new customers

We launched this feature last month, and given its favorable performance, now made it the default for all new accounts.

<Tip>Enabling transparent cache can speed up runs by 4x, without making any changes to your Github Actions caches.</Tip>

### **Faster runner provisioning times via improved pools**

By optimizing the types of runners we pre-provision and keep ready for incoming jobs, we've cut down average runner provisioning times by a further 2+ seconds.

### Github Actions caches can now be accessible from all scopes of your Github Repos

This is not a common usage pattern, but can be helpful for certain projects.

<Note>This feature is behind a feature flag, please open a ticket to enable.</Note>

<PrList ids={[2345, 2651]} />

## VMs and Networking

### Simplified Load Balancer port mapping for private networking

Clients now just need to open up the private port for private subnet range.<img src="https://mintcdn.com/ubicloud-00f14c0d/dkQS7vFTPxPT-89t/changelog/2025-01-private-subnet-console.png?fit=max&auto=format&n=dkQS7vFTPxPT-89t&q=85&s=55051cb80237a8f807dae8de5fa37df7" alt="" width="2762" height="1698" data-path="changelog/2025-01-private-subnet-console.png" />

### Load balancer stack (ipv4, ipv6 or dual) is selectable

This selection is available in both the UI and the API

<img src="https://mintcdn.com/ubicloud-00f14c0d/dkQS7vFTPxPT-89t/changelog/2025-01-load-balancer-dual-stack.png?fit=max&auto=format&n=dkQS7vFTPxPT-89t&q=85&s=25ee0b54aee66f31c622deed723bdac0" alt="" width="2484" height="1130" data-path="changelog/2025-01-load-balancer-dual-stack.png" />

<PrList ids={[2543, 2531, 2548, 2541]} />

## Bug fixes / maintenance

Multiple bug fixes and minor improvements, including:

* Demo docker image updated (Github demo was broken)
* Race condition fixed during VM provisioning for private networking
* Autofocus on OTP input field. Thank you @sh0n0 for the fix!

<PrList ids={[2586, 2544, 2605]} />