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

# Choose models and reasoning effort

> Understand model availability, use Fable locally, and select the reasoning level supported by each harness.

Sparkles exposes the models accepted by the active local harness. It does not grant model access or bypass the harness provider's authentication, account entitlements, rollout rules, or installed version.

## Why a model may be unavailable

The `/model` list depends on four things:

1. The selected harness is installed and available on `PATH`.
2. You are authenticated with that harness or its configured provider.
3. Your provider account can use the model.
4. The installed harness reports or accepts that model identifier.

Sparkles combines its built-in catalog with the effective current model reported by Claude Code, Codex, or OpenCode. A custom configured model can therefore appear even when it is not in the built-in list.

Run `sparkles doctor` to inspect the installed runtimes, then use `/agent` before `/model` when you need to switch providers.

## Use Fable

With Claude Code installed and authenticated:

```bash theme={"theme":{"light":"github-light","dark":"github-dark"}}
sparkles --runtime claude --model claude-fable-5 --effort ultracode
```

Inside the TUI, the same setup is:

1. Run `/agent` and choose `claude`.
2. Run `/model` and choose **Fable 5**.
3. Run `/effort` and choose **Ultracode**.

With an OpenCode provider that exposes Anthropic models:

```bash theme={"theme":{"light":"github-light","dark":"github-dark"}}
sparkles --runtime opencode --model anthropic/claude-fable-5 --effort max
```

OpenCode's Anthropic profile supports up to `max`; the Claude harness exposes the separate `ultracode` level for supported models.

## Built-in model catalog

The current catalog includes the following defaults. Installed-runtime discovery can add or preserve other configured models.

### Claude Code

| Label         | Model value             |
| ------------- | ----------------------- |
| Fable 5       | `claude-fable-5`        |
| Opus 5        | `opus`                  |
| Sonnet 5      | `sonnet`                |
| Opus 4.8 1M   | `claude-opus-4-8[1m]`   |
| Opus 4.8      | `claude-opus-4-8`       |
| Opus 4.7 1M   | `claude-opus-4-7[1m]`   |
| Opus 4.7      | `claude-opus-4-7`       |
| Opus 4.6 1M   | `claude-opus-4-6[1m]`   |
| Sonnet 4.6 1M | `claude-sonnet-4-6[1m]` |
| Sonnet 4.6    | `claude-sonnet-4-6`     |
| Haiku 4.5     | `haiku`                 |

Quote model values that contain square brackets so the shell does not expand them as a pathname pattern:

```bash theme={"theme":{"light":"github-light","dark":"github-dark"}}
sparkles --runtime claude --model 'claude-opus-4-8[1m]'
```

### Codex

| Label               | Model value           |
| ------------------- | --------------------- |
| GPT-6 Astra         | `gpt-6-astra`         |
| GPT-5.6 Sol         | `gpt-5.6-sol`         |
| GPT-5.6 Terra       | `gpt-5.6-terra`       |
| GPT-5.6 Luna        | `gpt-5.6-luna`        |
| GPT-5.5             | `gpt-5.5`             |
| GPT-5.4             | `gpt-5.4`             |
| GPT-5.4 mini        | `gpt-5.4-mini`        |
| GPT-5.3 Codex Spark | `gpt-5.3-codex-spark` |

### OpenCode

| Label         | Model value                     |
| ------------- | ------------------------------- |
| GLM 5.3       | `openrouter/z-ai/glm-5.3`       |
| Kimi K3       | `openrouter/moonshotai/kimi-k3` |
| Fable 5       | `anthropic/claude-fable-5`      |
| Opus 5        | `anthropic/claude-opus-5`       |
| Sonnet 5      | `anthropic/claude-sonnet-5`     |
| GPT-6 Astra   | `openai/gpt-6-astra`            |
| GPT-5.6 Sol   | `openai/gpt-5.6-sol`            |
| GPT-5.6 Terra | `openai/gpt-5.6-terra`          |
| GPT-5.6 Luna  | `openai/gpt-5.6-luna`           |
| GPT-5.5       | `openai/gpt-5.5`                |

## Reasoning effort values

Effort is model-specific. The TUI shows friendly labels while configuration and flags use the values below.

| Label      | Value       | Meaning                                              |
| ---------- | ----------- | ---------------------------------------------------- |
| None       | `none`      | No extended reasoning                                |
| Low        | `low`       | Answers sooner                                       |
| Medium     | `medium`    | Balanced reasoning                                   |
| High       | `high`      | Deeper reasoning                                     |
| Extra high | `xhigh`     | Deepest standard reasoning                           |
| Max        | `max`       | Maximum reasoning                                    |
| Ultra      | `ultra`     | Codex multi-agent reasoning on supported models      |
| Ultracode  | `ultracode` | Maximum agentic reasoning on supported Claude models |

The selector only offers values supported by the active model:

| Model family                                             | Available levels                |
| -------------------------------------------------------- | ------------------------------- |
| Claude Fable 5, Opus 5, Sonnet 5, Opus 4.7, and Opus 4.8 | Low through Max, plus Ultracode |
| Claude 4.6 models                                        | Low through Max                 |
| Claude Haiku 4.5                                         | No configurable effort          |
| Codex GPT-6 Astra                                        | Low through Max                 |
| Codex GPT-5.6 Sol and Terra                              | Low through Max, plus Ultra     |
| Codex GPT-5.6 Luna                                       | Low through Max                 |
| Other built-in Codex models                              | Low through Extra high          |
| OpenCode Anthropic models                                | Low through Max                 |
| OpenCode GPT-6 Astra                                     | Low through Max                 |
| Other OpenCode OpenAI models                             | None through Extra high         |
| OpenCode GLM 5.3                                         | Low, High, and Max              |
| OpenCode Kimi K3                                         | Max                             |

Use [CLI configuration](/cli/configuration) to make a harness, model, and effort the default for a repository.