Mercury 2.5

Sep 2026
Mercury 2.5
Compare in
on
$0.12per task?
$0.20 / $0.75per 1M in / outcache $0.02
781 tok/sfirst token in 3.8 s?
260Kcontext

Rankingin thinking mode?

Full ranking →
1Gemini 4 Argon77.52MiniMax M3.1 Flash72.53Sonnet 5.571.84Muse Wonder Full71.35Opus 5.569.5
···
94A.X K22.595Mercury 2.52.496Gemini 2.5 Pro2.2

Strong at

B
Reasoning#165 of 334Long context#111 of 273
C
Coding#75 of 132Agents & work tasks#122 of 175Knowledge#171 of 318

Alternatives?

ModelScoreCostPer taskWhat changes
CheaperMiMo v2.6 FlashthinkingOpen×27.2-47%$0.062Pass back reasoning_content+3BetterGPT-6 Lunalow×5.1-96%$0.005Function calling only with none · Chat Completions+3OpenDeepSeek V4.1 Flash 1017max×28.3+125%$0.26Pass back reasoning_content · Chat Completions+4CheaperGPT-5.6 Lunaxhigh×18.1-27%$0.085CheaperClaude Haiku 5.5high×14.3-32%$0.079Remove sampling parameters+5

History

2 versions
1.61.822.22.4Apr 26Jul 26Release dateAutomationBench-AA22.5
1.61.822.22.4Apr 26Jul 26Release dateAutomationBench-AA22.5

Benchmarks

in Agents & Tool Use
Terminal-Bench 4.00#75
GDPval-AA (work tasks) 2.10#122
AA-LCR (long context) 1.171.7#111
AutomationBench-AA2.4#95
AA-Briefcase 1.1 · Elo377.8#104

FAQ

How much does Mercury 2.5 cost?›

$0.20 per 1M input tokens and $0.75 per 1M output tokens; cached input $0.02 per 1M. A typical benchmark task costs about $0.12 at default effort.

Is Mercury 2.5 free to use?›

There is no free API; the cheapest paid price is $0.20 per 1M input tokens.

How good is Mercury 2.5, and what is it best at?›

#165 of 334 in Chat & Knowledge, #75 of 132 in Coding, #95 of 133 in Agents & Tool Use, #97 of 133 in Vision & Documents.

Is Mercury 2.5 better than Mercury 2?›

On AutomationBench-AA, Mercury 2.5 scores 2.4 vs 1.7 — about the same. It is 202% more expensive per task.

How can I use Mercury 2.5?›

Through the inception API.

What is the context window of Mercury 2.5?›

260K tokens, with up to 66K output tokens.

When was Mercury 2.5 released? Is there a newer version?›

Released 2026-09-08. It is the current version in its line.

Is Mercury 2.5 open source?›

No, Mercury 2.5 is a closed model available through an API. The closest open-weights option here is DeepSeek V4.1 Flash 1017.

Can Mercury 2.5 read images?›

No, it accepts text only.

What is Mercury 2.5?›

Mercury 2.5 is a closed language model from inception, released Sep 2026, the successor to Mercury 2.

What are the best alternatives to Mercury 2.5?›

MiMo v2.6 Flash (cheaper), GPT-6 Luna (better), DeepSeek V4.1 Flash 1017 (open weights).