As of: 2026-11-19 22:11 CET. 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-2-5/ # Mercury 2.5 Inception, Mercury family. Status: current. Type: reasoning model. Released: 2026-09-08. Training cutoff: not published. ## How to call it | Provider | Model id | Endpoint | Docs | |---|---|---|---| | Inception API (OpenAI-compatible) | `mercury-2.5` | https://api.inceptionlabs.ai/v1/chat/completions | https://docs.inceptionlabs.ai/get-started/models | | OpenRouter | `inception/mercury-2.5` | | https://openrouter.ai/inception/mercury-2.5 | | Baseten | | | https://www.baseten.co/ | ## Pricing - Input: $0.20 per 1M tokens - Output: $0.75 per 1M tokens - Unit as published: per 1M tokens (list price; 80% launch discount brings it to $0.04 / $0.15, end date not stated) - Source: https://docs.inceptionlabs.ai/get-started/models - Last checked: 2026-09-30 ## Specs - Input: text - Output: text - Context window: 260,000 tokens - Max output: 65,536 tokens - Open weights: no - Licence: proprietary ## What it has not seen - No published training cutoff. Counted from its release (2026-09-08): at least 1,680 AI events since (373 major, 65 historic), as of 2026-11-19. - Everything since August 2026: https://postcutoff.com/since/2026-08/ ## What stands out - Diffusion-based text generation at very high speed: A diffusion LLM (dLLM) that refines many tokens in parallel rather than one at a time; Inception claims 1,107 tokens/s on NVIDIA GPUs, and Artificial Analysis measured ~661 tokens/s. Source: https://www.inceptionlabs.ai/blog/introducing-mercury-2-5 - Fast reasoning with tool calling: Configurable reasoning effort, tool calling and structured outputs; Inception claims a 40% intelligence gain over Mercury 2 and places it near GPT-5.6 Luna (low), Gemini 3.5 Flash-Lite and Claude Haiku 4.5. Source: https://www.inceptionlabs.ai/blog/introducing-mercury-2-5 ## Notes Artificial Analysis lists the release as Sept 8, 2026, measures ~661 tok/s and gives an Intelligence Index of 12, below average for its price tier. It lists $0.25/$0.75, while Inception's docs list $0.20/$0.75 before the discount. Inception's blog page showed Sept 29, 2026 when fetched, and OpenRouter also has a separate `inception/mercury-2.5-preview` listing, so the Sept 8 date may be the preview. Speed figures are vendor claims without disclosed batch or hardware details (RuntimeWire). Customer claims: Augment Code context compaction 150 s → 27 s; OpenCall ~170 ms median latency. 100M free API tokens for new accounts. Mercury 2.5 is aimed at latency-sensitive work (voice agents, autocomplete, agent sub-calls) where speed matters more than peak intelligence. Sources: [Inception blog](https://www.inceptionlabs.ai/blog/introducing-mercury-2-5) · [Artificial Analysis](https://artificialanalysis.ai/models/mercury-2-5) · [RuntimeWire](https://runtimewire.com/article/inception-mercury-2-5-diffusion-language-model-launch) · [Inception on X](https://x.com/_inception_ai/status/2097365772289151417) Model page: https://postcutoff.com/m/mercury-2-5/