Models
The models available on each Perch plan, and how Roost chooses between them.
Perch runs on a range of models rather than a single one. Roost picks between them for you on every task, and you can pin a specific model by hand: Starter inside the included pool, Pro across the full registry. This page lists what each plan includes.
This list evolves
The rates below are the published list price for each model, per million tokens: input, output, and cached input (prompt-cache hits). Perch meters your usage against these rates and does not mark them up. Cached input is billed only for tokens the provider reuses from cache; everything else bills at the input rate. Published rates change from time to time; we reflect changes here. Rates shown are current as of August 2026.
Promo rates through September 1, 2026
GLM 5.2 and Kimi K2.6 / Kimi K2.7 Code show prior list struck through with the promo price beside it. Metering uses the promo price through September 1, 2026, then returns to prior list. MiniMax and everything else are unchanged.
Starter
The included pool of capable hosted models. Roost picks from it for you on every task, and you can also hand-pick any model in the pool by name on Perch Desktop and the CLI. Starter runs the Roost Standard and Standard Max tiers, and includes up to $20 of hosted usage each month. The premium models and the Pro and Pro Max tiers stay on Pro.
The Starter pool and its published rates:
| Model | Input / 1M | Output / 1M | Cache / 1M |
|---|---|---|---|
| Qwen 3.6 | $0.25 | $1.25 | $0.05 |
| DeepSeek V4 Flash | $0.14 | $0.28 | $0.0028 |
| Kimi K2.5 | $0.60 | $3.00 | $0.10 |
| GLM 5 | $1.00 | $3.20 | $0.20 |
| Qwen3 Coder | $0.45 | $1.80 | $0.09 |
| Nemotron Super | $0.15 | $0.65 | $0.03 |
| MiniMax M2 | $0.30 | $1.20 | $0.03 |
| Gemma 4 E2B | $0.04 | $0.08 | $0.04 |
| Gemma 4 31B | $0.14 | $0.40 | $0.14 |
Pro
Every model in the registry, left to Roost or pinned by hand, across all four Roost tiers (Standard, Standard Max, Pro, Pro Max). Pro includes the full Starter pool plus the premium models below, and up to $150/mo of total usage: unlimited Roost fair use plus up to $75 of manual model selection.
The premium models Pro adds on top of the Starter pool, with their published rates:
| Model | Input / 1M | Output / 1M | Cache / 1M |
|---|---|---|---|
| GLM 5.2 | |||
| DeepSeek V4 Pro | $1.74 | $3.48 | $0.14 |
| Kimi K2.6 | |||
| Kimi K2.7 Code | |||
| MiniMax M3 | $0.29 | $1.20 | $0.06 |
| Nemotron Ultra | $0.75 | $2.75 | $0.15 |
| Nemotron 3.5 Lightning | $0.10 | $0.25 | $0.05 |
| Grok 4.3 | $1.25 | $2.50 | $0.20 |
| Qwen 3.7 Plus | $0.40 | $1.60 | $0.08 |
| Inkling | $1.00 | $4.05 | $0.17 |
Bring your own key
Available on every plan, including the free one. Add a key from any OpenAI-compatible provider, or point Perch at your own endpoint and name any model you want. Perch usage limits do not apply to inference you pay for yourself, so your included allowance stays untouched while you use it. A cloud key works on Web, Desktop, and CLI; a model running on your own machine is CLI only. See Bring your own key.
At a glance
| Starter | Pro | |
|---|---|---|
| Model selection | Included pool, automatic or pinned | Every model, automatic or pinned |
| Premium models | No | Yes |
| Manual model pin | Included pool only | Any model |
| Roost tiers | Standard, Standard Max | All four (adds Pro, Pro Max) |
| Included usage | Up to $20 / month | Up to $150 / month (unlimited Roost + $75 manual) |
| Bring your own key | Yes | Yes |
| Images and scans on any model | Yes | Yes |
Model choice is not only for the model you talk to. On Perch Desktop you can also set which model the helpers Perch spawns run on, which is how you have one model check another's work. See Sub-agents.
For prices and plans, see Pricing. For how the routing works, see Perch Roost.
Images and scans
Every model in Perch can work with images, including the ones that cannot see. Share a screenshot, a scanned invoice, or a photo and Perch reads it for the model doing your task.
MCP servers
Perch connects to external data sources through the Model Context Protocol. It ships with legal research and browser automation servers out of the box, and you can add your own.