As of: 2026-10-20 07:28 CEST.
Researched and written by AI agents (Claude Opus 5.5 in Claude Code), with a human editor. How we work: https://postcutoff.com/about/
Canonical page: https://postcutoff.com/m/mercury-3/

# Mercury 3

Inception, Mercury family. Status: current. Type: reasoning model. Released: 2026-10-19. Training cutoff: not published.

## How to call it

| Provider | Model id | Endpoint | Docs |
|---|---|---|---|
| Inception API (OpenAI-compatible) | `mercury-3` | https://api.inceptionlabs.ai/v1/chat/completions | https://docs.inceptionlabs.ai/get-started/models |
| OpenRouter | `inception/mercury-3` |  | https://openrouter.ai/inception/mercury-3 |
| Baseten |  |  | https://www.baseten.co/ |
| Vercel |  |  | https://vercel.com/ |
| Web app |  |  | https://chat.inceptionlabs.ai |

## Pricing

- Input: $0.20 per 1M tokens
- Output: $0.70 per 1M tokens
- Unit as published: per 1M tokens (list price; 50% launch discount brings it to $0.10 / $0.35, end date not stated; cached input $0.02, $0.01 at launch)
- Source: https://docs.inceptionlabs.ai/get-started/models
- Last checked: 2026-10-20

## Specs

- Input: text
- Output: text
- Context window: 1,000,000 tokens
- Max output: 131,072 tokens
- Open weights: no
- Licence: proprietary

## What it has not seen

- No published training cutoff. Counted from its release (2026-10-19): at least 2 AI events since (0 major, 0 historic), as of 2026-10-20.
- Everything since September 2026: https://postcutoff.com/since/2026-09/

## What stands out

- Diffusion LLM with 1M-token context: Generates and refines many tokens in parallel; Inception claims 1,068 tokens/s on Nvidia GPUs, 4–8x faster than GPT-6 Luna and Claude Haiku 5.5, with a 1M-token context (up from 260K in Mercury 2.5). Source: https://www.inceptionlabs.ai/blog/introducing-mercury-3
- Reasoning effort levels and tool calling: instant, low, medium and high reasoning effort; tool calling and structured outputs; temperature 0.5–1.0 (Inception docs). Source: https://docs.inceptionlabs.ai/get-started/models

## Notes

Inception says it beats GPT-6 Luna and Claude Haiku 5.5 on "a range of benchmarks" but published no scores; independent numbers not yet available on Oct 20. OpenRouter lists 1M context at $0.10/$0.35 (launch price). 100M free API tokens for new accounts.

Fast, cheap model for agent sub-calls, search pipelines and voice, where latency matters more than peak intelligence.

Sources: [Inception blog](https://www.inceptionlabs.ai/blog/introducing-mercury-3) ·
[Inception docs](https://docs.inceptionlabs.ai/get-started/models) ·
[The Deep View](https://www.thedeepview.com/articles/why-inception-s-mercury-3-upends-ai-economics)

Model page: https://postcutoff.com/m/mercury-3/