[ for projects big and small ]
Plans and pricing
No free plan. Student is $9.99. Pro is $12 — under Perplexity Pro at $20. The membership is the savings; included credit is a small taste, then warehouse-rate tokens (open-weight included) and prepaid GPUs. The first $20 top-up adds $5 of open-weight bonus.
Plans for one person
Student $9.99 and Pro $12 are the membership. A small inference taste is included; the rest is warehouse-rate tokens and GPUs — open-weight included.
| Features | Best value Student$9.99Per month | Most popular Pro$12Per month | Ultra$45Per month |
|---|---|---|---|
| Core | |||
| Included inference credit | $2 / month | $4 / month | $12 / month |
| Seats | 1 | 1 | 1 |
| API rate limits | 2× | 3× | 6× |
| API keys | 5 | 10 | 25 |
| Dedicated deployments | 1 | 2 | 5 |
| Open-weight + closed inference | Included | Included | Included |
| Priority request queue | Not included | Included | Included |
| Pay-as-you-go tokens & GPUs | Included | Included | Included |
| OpenDoor Chat | |||
| AI app early access | Early access Coming soon | Early access Coming soon | Priority Coming soon |
| Included Chat messages | 20 / 5h · 60/wk | 45 / 5h · 135/wk | 100 / 5h · 300/wk |
| Support | |||
| Request logs & playground | Included | Included | Included |
| Agents add-on | $20 / month | $20 / month | $20 / month |
| OpenDoor Search | $0.10 / query or $20/mo | $0.10 / query or $20/mo | $0.10 / query or $20/mo |
| Get Student | Get Pro | Get Ultra | |
Membership is the profit; usage is the warehouse price. Included credit is a small taste — never a $50 giveaway. Open-weight and closed models share that stipend. Top up $20+ once and we add $5 of open-weight bonus (expires in 30 days). Tokens and GPU-seconds beyond the taste stay prepaid.
Add-on
Agents
Hosted OpenClaw, Hermes, NemoClaw, and OpenBot on your workspace. Included on Enterprise. Agent tokens still bill the same prepaid quota.
$20/month
Metered
OpenDoor Search
Visible on Tools and this page. Queries debit credits at list price — enable to use. The $20/month Web Search add-on covers Search for the month. Included on Enterprise. Site admins are unlimited. No third-party search keys.
$0.10/ query
Top up to start
No free credit at signup. Top up $20 or more and get $5 extra on open-weight models.
Pay as you go
Published $ / 1M input, cached input, and output. Batch is 50% off.
Choose your model
Serverless catalog first. Dedicated L4 / A100 / H100 when you need isolation.
Save 50% with batch processing via POST /v1/batches.
Serverless chat
USD per 1M tokens. Cached input is 50% of input until prompt cache ships.
| Model | Input | Cached input | Output | Tag |
|---|---|---|---|---|
deepseek-v3 Together (serverless) | $1.44 | $0.719 | $1.44 | Serverless |
llama-3.1-70b-instruct Together (serverless) | $1.01 | $0.506 | $1.01 | Serverless |
llama-3.1-8b-instruct Together (serverless) | $0.207 | $0.103 | $0.207 | Serverless |
mistral-7b-instruct Together (serverless) | $0.230 | $0.115 | $0.230 | Serverless |
qwen2.5-72b-instruct Together (serverless) | $1.38 | $0.690 | $1.38 | Serverless |
qwen2.5-7b-instruct Together (serverless) | $0.345 | $0.172 | $0.345 | Serverless |
qwen3.8-max Qwen | $2.20 | $1.10 | $6.60 | Serverless |
qwen3-coder-480b-a35b-instruct Vertex AI (Model Garden) | $0.165 | $0.083 | $1.32 | Serverless |
qwen3-next-80b-instruct Vertex AI (Model Garden) | $0.165 | $0.083 | $1.32 | Serverless |
qwen3-next-80b-thinking Vertex AI (Model Garden) | $0.165 | $0.083 | $1.32 | Serverless |
All pricing is per million tokens.
Embeddings
USD per 1M input tokens via POST /v1/embeddings.
| Model | Input / 1M | Provider |
|---|---|---|
| BAAI/bge-base-en-v1.5 | $0.0092 | Together (serverless) |
On-demand GPUs
Dedicated capacity. This Mac (Metal) is $0. GCP is list price — about 25% over Google’s all-in cost — billed per second while warm. Scale-to-zero is $0 idle.
- Cached input is billed at 50% of input until prompt-cache affinity ships.
- Serverless models need no GPU deploy step (wholesale warm path).
- On-demand GPUs map to dedicated capacity, not serverless.
Rates as of 2026-08-22T23:09:59.738Z · source: pricing_rules
See how a request is billed
Auth, policy, routing, then metering — the same path for serverless and dedicated.