---
title: "Models and speed"
description: "The model lineup, reasoning effort, speed tiers, and credential routing"
canonical_url: "https://hoplite.sh/docs/agent/models"
markdown_url: "https://hoplite.sh/docs/agent/models.md"
---

# Models and speed
URL: /docs/agent/models
LLM index: /llms.txt
Description: The model lineup, reasoning effort, speed tiers, and credential routing
Related: /docs/workspace/billing, /docs/agent/slash-commands, /docs/threads/cost

# Models and speed

Pick a model per thread from the new-thread dialog, switch mid-thread with [`/model`](/docs/agent/slash-commands), or set a project-wide default under **Settings → Project → Agent**.

## Platform models

These are billed to your workspace's [credit ledger](/docs/workspace/billing):

| Model | Provider | Context window | Reasoning effort | Fast tier |
| --- | --- | --- | --- | --- |
| **GPT-5.6 Sol** | OpenAI | 200k | none → max | yes |
| **GPT-5.6 Terra** (default) | OpenAI | 200k | none → max | yes |
| **GPT-5.6 Luna** | OpenAI | 200k | none → max | yes |
| **GPT-5.5** | OpenAI | 200k | none, low → xhigh | yes |
| **GPT-5.3 Codex** | OpenAI | 400k | low → high | no |
| **Fable 5.1** | Anthropic | 1M | low → max | no |
| **Fable 5** | Anthropic | 200k | none → xhigh | no |
| **Opus 5** | Anthropic | 1M | none, low → max | no |
| **Opus 4.8** | Anthropic | 200k | none → xhigh | no |
| **Sonnet 5** | Anthropic | 1M | none, low → max | no |
| **Haiku 4.5** | Anthropic | 200k | none → xhigh | no |
| **Muse Spark 1.3** | Meta via OpenRouter | 1,048,576 | minimal → xhigh | no |

All platform models accept image input, so screenshots and design files in your messages reach the model directly.

## Reasoning effort

Reasoning effort trades latency and cost for depth: `none`, `minimal`, `low`, `medium` (the default), `high`, `xhigh`, and `max`. The picker only shows the levels the selected model actually supports. Reasoning tokens are billed as output tokens.

## Speed

Models with a fast tier (currently the GPT family) offer a **standard / fast** toggle. Fast maps to a priority service tier — same model, quicker output.

## Automatic ChatGPT / API routing

There is one model list — no separate `codex/` copies. If you connect ChatGPT under **Settings → Account → Profile → Connect ChatGPT**, supported OpenAI models automatically use that subscription and are **billed to your ChatGPT plan, not your workspace credits**. If no eligible connected subscription is available for the thread, Hoplite runs the same selected model through the platform API instead. Usage is still recorded on the [usage page](/docs/workspace/billing), with the actual payer tracked per call.

## Claude Code subscription routing

Workspaces enrolled in the Claude Code rollout show **Claude** under **Settings
→ Integrations → Personal**. Run `claude setup-token` on your own machine and
paste the generated token to connect your Claude subscription. Hoplite stores
the token encrypted and never returns it after saving.

For organizations enrolled in the Claude Code rollout, eligible cloud runs on supported Anthropic models can use the initiating member's connected subscription. Hoplite runs Claude Code in an isolated, worker-owned Modal ACP environment and gives it an authenticated Hoplite MCP bridge to the thread workspace, so the subscription token never enters the tenant sandbox. If the organization, member, or model is not eligible, Hoplite keeps the ordinary platform or BYOK route instead.

## Bring your own provider keys

When BYOK is enabled for a workspace, its owners and admins can manage direct Anthropic and OpenAI API keys under **Settings → Workspace → General**. Keys are encrypted, are never displayed after saving, and can be replaced or removed at any time.

Compatible runs use the workspace key for their selected provider; when that provider has no configured workspace key, Hoplite falls back to its platform credentials. This workspace-level provider routing is separate from a person's ChatGPT subscription connection. See [Cost](/docs/threads/cost) for how BYOK runs are badged in the cost breakdown.

## Housekeeping models

Small, fast delegate models handle background work like generating thread titles and context compaction — you don't pick these, and they keep housekeeping costs negligible.

## Sitemap

See the full [sitemap](/docs/sitemap.md) for all pages.
Well-known sitemap: [/docs/.well-known/sitemap.md](/docs/.well-known/sitemap.md).