Atlas

Models

← All models

Mistral Large 2.1 (Instruct 2411)

Mistral AI · 2024-11-18 · 122.6B parameters

Mistral Large 2.1 is Mistral AI's Mistral-Large-Instruct-2411 checkpoint, released on November 18, 2024 as an update to the July 2407 artifact. The 123B-class dense model improved long-context understanding, system-prompt handling, and function calling while supporting a 128K-token context window. Mistral retired the hosted `mistral-large-2411` snapshot on May 31, 2026 and lists Mistral Medium 3.5 as its replacement.

Benchmark scores

BenchmarkScore
A-Fantasia78.3
A-Fantasia Backwards Spelling98.0
A-Fantasia Chess62.0
A-Fantasia Cube Rotation75.0
AA-LCR5.3
AA-Omniscience Index-34.0
AIME 2024 (avg@8)11.7
AIME 2024-20259.2
AIME 202514.0
AIME 2025 (avg@8)6.7
Arena Text — Business, Management & Financial Ops (No Style Control)1238
Arena Text — Chinese (No Style Control)1239
Arena Text — Creative Writing (No Style Control)1242
Arena Text — English (No Style Control)1284
Arena Text — Entertainment, Sports & Media (No Style Control)1231
Arena Text — Exclude Ties (No Style Control)1180
Arena Text — Exclude Ties (Style Controlled)1237
Arena Text — Expert (No Style Control)1207
Arena Text — French (No Style Control)1324
Arena Text — French (Style Controlled)1373
Arena Text — German (No Style Control)1242
Arena Text — Hard Prompts (No Style Control)1257
Arena Text — Hard Prompts English (No Style Control)1269
Arena Text — Instruction Following (No Style Control)1249
Arena Text — Japanese (No Style Control)1171
Arena Text — Korean (No Style Control)1202
Arena Text — Legal & Government (No Style Control)1275
Arena Text — Life, Physical & Social Science (No Style Control)1261
Arena Text — Longer Query (No Style Control)1262
Arena Text — Medicine & Healthcare (No Style Control)1254
Loading Atlas data…