Model specification · Mistral

Mistral Large 3 2512

mistralai/mistral-large-2512

Mistral Large 3 2512 is Mistral’s most capable model to date, featuring a sparse mixture-of-experts architecture with 41B active parameters (675B total), and released under the Apache 2.0 license.

SpecificationsReported values
Context window
262K
Maximum output
210K
Architecture
text+image+file->text
Tokenizer
Mistral
Knowledge cutoff
—
Moderated
No
Input modalities
textimagefile
Output modalities
text
Full pricingUSD where reported
PriceStandard
Input$0.500 / M
Output$1.50 / M
Benchmark indexesReported indexes
Intelligence
9.3
Coding
20.1
Agentic
1
Design Arena Elo
1099
Example workload costsStandard pricing only
WorkloadInput / outputCache shareEstimated cost
Quick check10K / 1K × 10%$0.00650
Long analysis100K / 4K × 150%$0.034
Workday20K / 2K × 1,00030%$10.30
More from MistralAll Mistral models
ModelContextInput / MOutput / MIntel.
Mistral Large 41.0M$0.680$2.0938.4
Mistral Medium 3.5262K$1.50$7.5014.2
Mistral Small 4262K$0.150$0.60011.3
Mistral Medium 3.1131K$0.400$2.009.2
Mistral Medium 3131K$0.400$2.009
Devstral 2 2512262K$0.400$2.008.6
Mistral Large 2407131K$2.00$6.007.6
Mistral Small 333K$0.050$0.0806.7
Saba33K$0.200$0.6006.5
Ministral 3 14B 2512262K$0.200$0.2006
Mistral Large128K$2.00$6.005.8
Mixtral 8x22B Instruct66K$2.00$6.005.7