Inception Labs / inception/mercury-3

Mercury 3 - access through LLMTR

Mercury 3 is Inception Labs' diffusion language model (dLLM): it generates several tokens in parallel and refines them, for low-latency agents, voice and high-volume automation. It has a 1,000,000-token context window and a single response can be at most 131,072 tokens. It supports tool calling and schema-enforced JSON output via `response_format`, and accepts text only. Repeated prompt prefixes can be served from a prompt cache at one tenth of the input rate. It reasons by default at the `low` level and reasoning tokens bill as output; the levels are `:minimal`, `:low`, `:medium` and `:high`, and `:minimal` gives the lowest latency. The introductory price applies until the end of 25 October 2026 (TRT); from 26 October 2026 the standard rate applies.

Technical specifications

Canonical IDinception/mercury-3
ProviderInception Labs
Context window1,000,000 tokens
OperationsCHAT_COMPLETIONS
Modalitiestext

Pricing

An 8% platform margin applies to credit top-ups; model usage prices are not separately marked up.

OperationMetricUnitPrice
CHAT_COMPLETIONSINPUT_TEXTPER_1M_TOKENS$0.100000
CHAT_COMPLETIONSCACHE_READPER_1M_TOKENS$0.010000
CHAT_COMPLETIONSOUTPUT_TEXTPER_1M_TOKENS$0.350000

Example usage

With existing OpenAI SDK flows, change only the base URL and model identifier.

curl https://llmtr.com/v1/chat/completions   -H "Authorization: Bearer llmtr-your_key"   -H "Content-Type: application/json"   -d '{"model":"inception/mercury-3","messages":[{"role":"user","content":"Hello"}]}'

Guides about this model

Related models