Mercury 2

Feb 2026
Mercury 2
Compare in
on
$0.039per task?
$0.25 / $0.75per 1M in / outcache $0.025
487 tok/sfirst token in 8.3 s?
128Kcontext

Rankingat high effort?

Strong at

A
Instructions#52 of 245
B
Tool calling#92 of 242Reasoning#135 of 334Science#134 of 312
C
Coding#75 of 132Agents & work tasks#122 of 175Long context#188 of 273Knowledge#209 of 318

Alternatives?

ModelScoreCostPer taskWhat changes
CheaperAgnes 2.5 Pro Beta×3.3-60%$0.10 / $0.30BetterGPT-6 Lunalow+100%-88%$0.005Function calling only with none · Chat Completions+3OpenMiMo v2.6 Prothinking×4.3+245%$0.13Pass back reasoning_content+3CheaperClaude Haiku 5.5low×2.7-37%$0.024Remove sampling parameters+5CheaperGPT-5.6 Lunamedium×2.3-60%$0.016

History

2 versions
10.51111.51212.5Apr 26Jul 26Release dateAA Intelligence Index 4.3.222.5
10.51111.51212.5Apr 26Jul 26Release dateAA Intelligence Index 4.3.222.5

Benchmarks

all categories
Chat & Knowledge
Humanity's Last Exam17.1#135
AA Omniscience Index-50.7#210
CritPt (physics)0.8#140
AA Intelligence Index 4.3.210.8#202
AA Omniscience · Accuracy21.2#142
AA Omniscience · Hallucination Rate91.2#264
AA-LCR (long context) 1.143.7#190
GDP.pdf · AA all-pass3.8#94
GPQA Diamond77#135
Coding
Terminal-Bench 4.00#75
SciCode37.7#108
Terminal-Bench Hard (version unstated)26.5#68
Terminal-Bench 2.127.3#102
Agents & Tool Use
AutomationBench-AA1.7#102
GDPval-AA (work tasks) 2.10#122
τ³-Banking · AA9.5#94
AA-Briefcase 1.1 · Elo265.3#111
τ²-bench · AA70.8#94
Writing & Language
IFBench (instructions)69.8#52

FAQ

How much does Mercury 2 cost?›

$0.25 per 1M input tokens and $0.75 per 1M output tokens; cached input $0.025 per 1M. A typical benchmark task costs about $0.039 at high effort.

Is Mercury 2 free to use?›

There is no free API; the cheapest paid price is $0.25 per 1M input tokens.

How good is Mercury 2, and what is it best at?›

#135 of 334 in Chat & Knowledge, #75 of 132 in Coding, #102 of 133 in Agents & Tool Use, #134 of 195 in Writing & Language, #94 of 133 in Vision & Documents. Strongest at: Instructions.

Is Mercury 2 better than Trinity Large Thinking?›

On AA Intelligence Index 4.3.2, Mercury 2 scores 10.8 vs 10.8 — about the same. It is 67% cheaper per task.

How can I use Mercury 2?›

Through the API of Inception.

What is the context window of Mercury 2?›

128K tokens, with up to 50K output tokens.

When was Mercury 2 released? Is there a newer version?›

Released 2026-02-20. It has been superseded by Mercury 2.5.

Is Mercury 2 open source?›

No, Mercury 2 is a closed model available through an API. The closest open-weights option here is MiMo v2.6 Pro.

Can Mercury 2 read images?›

No, it accepts text only.

What is Mercury 2?›

Mercury 2 is a closed language model from inception, released Feb 2026.

What are the best alternatives to Mercury 2?›

Agnes 2.5 Pro Beta (cheaper), GPT-6 Luna (better), MiMo v2.6 Pro (open weights).