docs(console): every harness reaches every model (#59611)

This commit is contained in:
Victor Navarro
2026-10-27 13:56:20 +01:00
committed by GitHub
Unverified
parent 86976449c4
commit 0b1f1251be
5 changed files with 31 additions and 67 deletions
+3 -1
View File
@@ -371,7 +371,9 @@ For these models, you still get a little more than if you paid the model provide
## Endpoints
You can also access Go models through the following API endpoints.
You can also access Go models through the following API endpoints. Each model is listed with its native API, but every
text model also works through the other APIs, which Console translates. See
[Base URLs](/console/harnesses#base-urls) for the full set.
<div class="docs-table-scroll" role="region" aria-label="Go model endpoints" tabIndex={0}>
@@ -3,8 +3,9 @@ title: "Claude Code"
description: "Use Console and Go models from Claude Code."
---
Claude Code speaks the Anthropic Messages API. Point it at Console with two environment variables and use any model
that accepts Anthropic Messages.
Claude Code speaks the Anthropic Messages API. Point it at Console with two environment variables and use any paid or Go
model, including GPT, Gemini, and open models. Console translates the request for models that don't accept Anthropic
Messages.
```bash
export ANTHROPIC_BASE_URL="https://opencode.ai/inference/anthropic"
@@ -56,29 +57,11 @@ Use the Go base URL and a key from a workspace with a [Go](/console/go) subscrip
## Models
Claude Code can only use models that accept Anthropic Messages.
Use any model ID from the [Console](/console/models#endpoints) or [Go](/console/go#endpoints) table. Switch models in a
session with `/model`, or pick one for a single run.
| Plan | Models |
| ------------- | ------------------------------------------------------------------- |
| Pay-as-you-go | Claude models and most Qwen models |
| Go | Claude Haiku 5.5, Kimi K3, DeepSeek V4, MiniMax M3, and Qwen models |
GPT, Gemini, and most open models are not available in Claude Code yet. Use [Pi](/console/harnesses/pi) or
[Hermes](/console/harnesses/hermes) for them in the meantime.
<Callout>Support for every model in every harness is coming soon.</Callout>
### Non-Claude models
Claude Code sends Claude-specific request options to every model. If a non-Claude model rejects a request with a
`400` error, turn off experimental betas.
```json title="~/.claude/settings.json"
{
"env": {
"CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS": "1"
}
}
```bash
claude --model gpt-6.1-sol
```
## Console MCP
@@ -3,8 +3,8 @@ title: "Codex"
description: "Use Console and Go models from Codex."
---
Codex speaks the OpenAI Responses API, so it can use the models served through Responses, such as GPT, Grok, and Muse
Spark.
Codex speaks the OpenAI Responses API. Point it at Console and use any paid or Go model, including Claude, Gemini, and
open models. Console translates the request for models that don't accept Responses.
## Configure
@@ -38,7 +38,7 @@ codex
Use the Go base URL and a key from a workspace with a [Go](/console/go) subscription.
```toml title="~/.codex/config.toml"
model = "gpt-6-luna"
model = "kimi-k3"
model_provider = "opencode-go"
[model_providers.opencode-go]
@@ -50,22 +50,13 @@ wire_api = "responses"
## Models
Set `model` to a model that accepts the Responses API. Override it for one run with `--model`.
| Plan | Models |
| ------------- | -------------------------------------- |
| Pay-as-you-go | GPT, Grok, and Muse Spark models |
| Go | GPT, Grok, Muse Spark, and DeepSeek V4 |
Set `model` to any model ID from the [Console](/console/models#endpoints) or [Go](/console/go#endpoints) table. Override
it for one run with `--model`.
```bash
codex --model grok-4.7
codex --model claude-sonnet-5-5
```
Claude, Gemini, and most open models are not available in Codex yet. Use [Pi](/console/harnesses/pi) or
[Hermes](/console/harnesses/hermes) for them in the meantime.
<Callout>Support for every model in every harness is coming soon.</Callout>
Codex doesn't know the context window or reasoning settings of models outside its own catalog. It uses generic defaults
for them and prints a warning, which you can ignore.
@@ -75,7 +66,7 @@ Keep pay-as-you-go and Go side by side with a profile. Define both providers in
profile file that switches to Go.
```toml title="~/.codex/go.config.toml"
model = "gpt-6-luna"
model = "kimi-k3"
model_provider = "opencode-go"
```
@@ -3,16 +3,16 @@ title: "Harnesses"
description: "Use Console and Go models from other coding agents and AI apps."
---
Console works with coding agents other than OpenCode. Point a harness at the Console gateway with an API key, and usage
is billed to your workspace like any other request.
Console works with coding agents other than OpenCode. Point a harness at the Console gateway with an API key to use any
paid or Go model, and usage is billed to your workspace like any other request.
| Harness | Pay-as-you-go | Go | Models | MCP |
| --------------------------------------------- | ------------- | --------- | ------------------------- | --- |
| [Claude Code](/console/harnesses/claude-code) | Yes | Yes | Anthropic Messages models | Yes |
| [Codex](/console/harnesses/codex) | Yes | Yes | OpenAI Responses models | Yes |
| [Hermes](/console/harnesses/hermes) | Built in | Built in | Paid and Go models | Yes |
| [Pi](/console/harnesses/pi) | Built in | Built in | Paid and Go models | Yes |
| [Raycast](/console/harnesses/raycast) | Extension | Extension | Paid and Go models | Yes |
| Harness | Pay-as-you-go | Go | MCP |
| --------------------------------------------- | ------------- | --------- | --- |
| [Claude Code](/console/harnesses/claude-code) | Yes | Yes | Yes |
| [Codex](/console/harnesses/codex) | Yes | Yes | Yes |
| [Hermes](/console/harnesses/hermes) | Built in | Built in | Yes |
| [Pi](/console/harnesses/pi) | Built in | Built in | Yes |
| [Raycast](/console/harnesses/raycast) | Extension | Extension | Yes |
Free-tier chat models are only available from within OpenCode. Other harnesses receive HTTP `403` for those models,
even with a Console API key. Use paid models or a Go subscription.
@@ -46,7 +46,8 @@ Pay-as-you-go and Go are served from different base URLs. Use the one that match
| Pay-as-you-go | `https://opencode.ai/inference` |
| Go | `https://opencode.ai/inference/go` |
Append the path for the API your harness speaks.
Append the path for the API your harness speaks. Every API serves every paid and Go text model: Console sends a request
to the model's own endpoint when it speaks that API, and translates it otherwise.
| API | Path |
| ----------------------- | ----------------------------------------------- |
@@ -59,22 +60,6 @@ Append the path for the API your harness speaks.
https://opencode.ai/inference/go/openai/v1/chat/completions
```
## Choose an API
A harness can only use the models that accept the API it speaks. Most models accept one API, and some accept several.
The endpoint column in the [Console](/console/models#endpoints) and [Go](/console/go#endpoints) tables lists one API per
model, so check the harness page for the full list.
Hermes, Pi, and the Raycast extension pick the right API for each paid or Go model. Claude Code and
Codex speak a single API and only reach part of the catalog.
| Harness | API | Models |
| ----------- | ------------------ | ---------------------------------------------------------------- |
| Claude Code | Anthropic Messages | Claude and Qwen, plus Kimi K3, DeepSeek V4, and MiniMax M3 on Go |
| Codex | OpenAI Responses | GPT, Grok, and Muse Spark, plus DeepSeek V4 on Go |
<Callout>Support for every model in every harness is coming soon.</Callout>
## Sessions
Send a stable session ID with every request so Console can route a conversation consistently and reuse its prompt
@@ -31,6 +31,9 @@ You are charged per request and can add credits to your account.
You can also access the models directly through the following API endpoints. Free-tier chat models require OpenCode;
use paid models or [Go](/console/go) from other clients.
Each model is listed with its native API, but every text model also works through the other APIs, which Console
translates. See [Base URLs](/console/harnesses#base-urls) for the full set.
<div class="docs-table-scroll" role="region" aria-label="Console model endpoints" tabIndex={0}>
| Model | Model ID | Endpoint | AI SDK Package |