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

# Model benchmarks

> Independent scores for the models Aura actually uses.

<Frame>
  <img src="https://chat.youniverse.ai/__l5e/assets-v1/a23392fb-f515-472d-9b0c-9d1acf13e7c6/docs-learn-bench-bars.jpg" alt="Glass bar chart topped with colored orbs" />
</Frame>

The models behind Aura's chat modes, with independent scores from **Artificial Analysis** (intelligence, coding and agentic indexes) and **Design Arena** (head-to-head human votes on generated apps, sites and 3D). Data as of October 2026.

## Where each model is used

| Model | Used in |
| - | - |
| **Claude Haiku 5.5** | Unlimited, Orchestration (Claude), Frontier Low |
| **Claude Opus 5.5** | Frontier Efficient and High, AI Builder |
| **GPT-6.1 Sol** | Orchestration (GPT), Frontier Efficient |
| **GPT-6 Astra** | Frontier High (GPT) |
| **GPT-6 Luna** | Frontier Low (GPT), fast background tasks |
| **Gemini 3.8 Flash** | Orchestration and Frontier (Gemini), Work Mode, Agents |
| **Grok 4.7** | Orchestration and Frontier (Grok) |

## Intelligence, coding and agentic indexes

Higher is better.

| Model | Intelligence | Coding | Agentic |
| - | - | - | - |
| Claude Opus 5.5 | **57.6** | — | — |
| GPT-6 Astra | 52.7 | **76.9** | **51.0** |
| GPT-6.1 Sol | 51.8 | — | — |
| Grok 4.7 | 46.4 | — | — |
| Claude Haiku 5.5 | 43.4 | — | — |
| Gemini 3.8 Flash | 40.9 | 76.3 | 40.2 |
| GPT-6 Luna | 38.1 | — | — |

"—" means that index hasn't been published for the model yet.

<Frame>
  <img src="https://chat.youniverse.ai/__l5e/assets-v1/c8825254-194d-4d6f-96f0-dbdc9eddfe13/docs-learn-bench-podium.jpg" alt="Glass podium with trophies and a radar chart" />
</Frame>

## Design Arena rankings (human votes)

Rank among all models tested in each category.

| Category | Opus 5.5 | GPT-6.1 Sol | Haiku 5.5 | Gemini 3.8 Flash | Grok 4.7 |
| - | - | - | - | - | - |
| 3D | **#1** | #2 | #6 | #16 | #23 |
| Coding | **#1** | #5 | #20 | #13 | #29 |
| Game dev | **#1** | #4 | #15 | #13 | #16 |
| UI components | **#1** | — | — | #9 | #24 |
| Data visualization | **#1** | #9 | — | #25 | #34 |
| Websites | #3 | #5 | #28 | #11 | #38 |

## Why Aura combines models

No single model wins everything. Opus leads on design and code, GPT-6 Astra on agentic work, Gemini 3.8 Flash on speed with native audio and video, Grok on live information. [Orchestration](/learn/how-orchestration-works) asks four of them and compares, so you get each model's strengths and catch their blind spots.

## Price per million tokens

| Model | Input | Output |
| - | - | - |
| Claude Haiku 5.5 | \$0.10 | \$0.50 |
| GPT-6 Luna | \$0.10 | \$0.50 |
| Gemini 3.8 Flash | \$0.75 | \$3.75 |
| GPT-6.1 Sol | \$2.00 | \$10.00 |
| Grok 4.7 | \$2.00 | \$6.00 |
| Claude Opus 5.5 | \$4.00 | \$20.00 |
| GPT-6 Astra | \$10.00 | \$50.00 |

<Note>
  Aura upgrades models as better versions ship; the app always shows the version currently answering.
</Note>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.