Instamodel
Manage keysTry public demo
Skip to comparison
COMPARE / ROUTE / RECEIPT

Put the right model
to the same test.

Run frontier and specialist models side by side. Every response keeps its provider route, token usage, cost, and latency attached.

Browse full catalog
03 / COMPARE

Response matrix

Execution modeLocal previews

Ready for the public deterministic demo and clearly labeled local previews.

A
Meta

Llama 4 Maverick

Open weights
1M context$0.22 in$0.88 out
ResponseLocal simulation

LOCAL SIMULATION — no request was sent to Llama 4 Maverick. This preview demonstrates the response surface only. A live request would route the prompt to meta-llama/llama-4-maverick, then return the provider's generated text with metered token, cost, and latency data.

Usage receiptsim_a_meta_llama_llama_4_maverick
Input
43
tokens
Output
148
tokens
Total
191
tokens
Cost
$0.0001
USD
Latency
427
ms
Served byFireworks AI

Local simulation; balanced preference would be recorded but does not switch providers in the private alpha

B
Northstar Legal

ClauseGuard 32B

Post-trained
128K context$0.32 in$0.78 out
ResponseLocal simulation

LOCAL SIMULATION — no request was sent to ClauseGuard 32B. This preview demonstrates the response surface only. A live request would route the prompt to northstar-demo/clauseguard-32b, then return the provider's generated text with metered token, cost, and latency data.

Usage receiptsim_b_northstar_demo_clauseguard_32b
Input
43
tokens
Output
112
tokens
Total
155
tokens
Cost
$0.0001
USD
Latency
373
ms
Served byBaseten

Local simulation; balanced preference would be recorded but does not switch providers in the private alpha

OPENAI-COMPATIBLEPOST /api/v1/chat/completions

Change the model ID, not your application. Routing preferences travel with the request and the receipt comes back with the response.