Mercury 2.5
Mercury 2.5 is Inception’s diffusion-based reasoning model for chat, agents, and structured workflows, with tool calling, structured outputs, and a 260K context window.
text → text · made by Inception Labs
Sold by 7 ways
| Seller | Lane | Rate |
|---|---|---|
| Nous Research aggregator | standard | $0.04per Mtok in$0.15per Mtok out$0.004per Mtok cachedinference-api.nousresearch.com · read 2026-09-20 |
| OpenRouter 2 ways | $0.04per Mtok in$0.15per Mtok out$0.004per Mtok cached | |
| OpenRouter aggregator | standard | $0.04per Mtok in$0.15per Mtok out$0.004per Mtok cachedopenrouter.ai · read 2026-09-02 |
| OpenRouter aggregator | inception | $0.04per Mtok in$0.15per Mtok out$0.004per Mtok cachedopenrouter.ai · read 2026-09-20 |
| Vercel AI Gateway aggregator | standard | $0.04per Mtok in$0.15per Mtok out$0.004per Mtok cachedai-gateway.vercel.sh · read 2026-09-20 |
| nano-gpt 0 ways | $0.04per Mtok in$0.15per Mtok out$0.004per Mtok cached | |
| nano-gpt aggregatornot on the seller's list today | standard | $0.04per Mtok in$0.15per Mtok out$0.004per Mtok cachedmodels.dev · read 2026-09-02 |
| kilo 0 ways | $0.2per Mtok in$0.75per Mtok out$0.02per Mtok cached | |
| kilo aggregatornot on the seller's list today | standard | $0.2per Mtok in$0.75per Mtok out$0.02per Mtok cachedmodels.dev · read 2026-09-02 |
| Venice AI aggregator | standard | $0.25per Mtok in$0.94per Mtok out$0.025per Mtok cachedmodels.dev · read 2026-10-02 |
Measured 3 standings
| Place | Board | Metric | Score |
|---|---|---|---|
| 69thof 73 | Proofbench — Epoch AI | Accuracy | 3 |
| 111thof 119 | Ale bench — Epoch AI | Performance (high) | 301.65 |
| 139thof 263 | Artificial Analysis · Intelligence Index | Artificial Analysis Intelligence Index | 12 |
About
Mercury 2.5 — a text model from Inception Labs, sold by 4 companies from $0.04 in and $0.15 out per million tokens, placed 139th of 263 on Artificial Analysis · Intelligence Index.
It takes text and returns text, with a context window of 260,000 tokens. It was published in September 2026, trained on material up to November 2025. Its sellers say it can reason step by step and call a tool. The catalogue files it under chat. Four companies sell it. The cheapest is $0.04 in and $0.15 out per million tokens at Nous Research. Beside the standard rate there is a separately routed lane. It has been measured on 3 boards, and stands best at 139th of 263 on Artificial Analysis · Intelligence Index.
Every current figure
- Maker
- Inception Labs
- Register
- model
- Takes
- text
- Returns
- text
- Context
- 260,000 tokens
- Longest answer
- 65,536 tokens
- Published
- September 2026
- Knowledge to
- November 2025
- Licence
- not read
- Sellers
- 4
- Maker's own price
- not read
- Price
- $0.04 in and $0.15 out per million tokens — Nous Research
- Boards
- 3
- Best place
- 139th of 263 — Artificial Analysis · Intelligence Index
Known as 1 name
inception/mercury-2.5