Providers

Together AI

Together AI provides access to leading open-source models including Llama, DeepSeek, Kimi, and more through a unified API. OpenClaw bundles it as the together provider.

Property	Value
Provider	`together` (alias: `togetherai`)
Auth	`TOGETHER_API_KEY`
API	OpenAI-compatible
Base URL	`https://api.together.xyz/v1`

Getting started

Get an API key

Create an API key at api.together.ai/settings/api-keys.

Run onboarding

bash

openclaw onboard --auth-choice together-api-key

Set a default model

json5

{  agents: {    defaults: {      model: {        primary: "together/zai-org/GLM-5.2",      },    },  },}

Non-interactive example

bash

openclaw onboard --non-interactive --accept-risk --skip-health \  --mode local \  --auth-choice together-api-key \  --together-api-key "$TOGETHER_API_KEY"

Built-in catalog

Cost is USD per million tokens.

Model ref	Name	Input	Context	Max output	Cost (in/out)	Notes
`together/meta-llama/Llama-3.3-70B-Instruct-Turbo`	Llama 3.3 70B Instruct Turbo	text	131,072	8,192	1.04 / 1.04	General model
`together/zai-org/GLM-5.2`	GLM 5.2 FP4	text	262,144	131,072	1.40 / 4.40	Default model

Video generation

The bundled together plugin also registers video generation through the shared video_generate tool.

Property	Value
Default video model	`Wan-AI/wan2.7-t2v`
Other models	`Wan-AI/wan2.7-i2v`, `minimax/hailuo-02`
Modes	text-to-video; image-to-video only with `Wan-AI/wan2.7-i2v` (single first-frame image)
Duration	Wan 2.7: 2-10 seconds in OpenClaw; Hailuo duration support is model-specific
Supported parameters	`size` (see below); `aspectRatio`/`resolution` are not read

For Wan 2.7, OpenClaw translates size into Together's resolution and ratio fields. Supported sizes follow Together's Wan 2.7 guide:

Aspect ratio	720P size	1080P size
16:9	`1280x720`	`1920x1080`
9:16	`720x1280`	`1080x1920`
1:1	`960x960`	`1440x1440`
4:3	`1104x832`	`1648x1248`
3:4	`832x1104`	`1248x1648`

Other Wan sizes are rejected before submission. Omit size to use Together's defaults. Hailuo continues to receive pixel width and height from size. Wan image input is sent as media.frame_images with frame: "first", not as a style reference in media.reference_images.

To use Together as the default video provider:

json5

{  agents: {    defaults: {      mediaModels: {        video: {          primary: "together/Wan-AI/wan2.7-t2v",        },      },    },  },}

Prompt caching

Together automatically reuses matching prompt prefixes. OpenClaw sends a stable prompt_cache_key on native Together requests to improve routing across turns. Cache usage is read from both usage.prompt_tokens_details.cached_tokens and usage.cached_tokens, depending on the model's response format. Keep the model, tool definitions, reasoning settings, and earlier messages stable for the best reuse. See Together prompt caching.

Environment note

If the Gateway runs as a daemon (launchd/systemd), make sure TOGETHER_API_KEY is available to that process (for example, in ~/.openclaw/.env or via env.shellEnv).

Troubleshooting

Verify your key works: openclaw models list --provider together
If models are not appearing, confirm the API key is set in the correct environment for your Gateway process.
Model refs use the form together/<model-id>.

Model providers

Provider rules, model refs, and failover behavior.

Video generation

Shared video generation tool parameters and provider selection.

Configuration reference

Full config schema including provider settings.

Together AI

Together AI dashboard, API docs, and pricing.

Was this useful?