BenchLeader
Menu
LeaderboardsModelsProvidersBenchmarksCompareReleasesBlogMethodology
Mistral AIOpen weightsAuto-detected

Mixtral 8x7B

Best configuration ranks #595 of 610 on the BenchLeader Index at 35.1 ±5.1. Last measured 2 Sept 2026. Released 11 Dec 2023.

Blended price
$0.513/M
$0.450 in · $0.700 out
Output speed
–
First answer
–
Context
33k
How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.
  1. Overall index35
  2. Reasoning28
  3. Coding35
  4. Maths28
  5. Knowledge17
  6. Human preference34
  7. Composite35

Benchmark results

One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.

Reasoning

BenchmarkdefaultSource
GPQA Diamond30.6%#250Epoch AI Benchmarking Hub
LMArena Hard Prompts1210#289LMArena
GPQA Diamond (AA)not in index29.2%#494Artificial Analysis
Humanity's Last Exam (AA)not in index4.7%#402Artificial Analysis

Coding

BenchmarkdefaultSource
LMArena Coding1239#288LMArena

Maths

BenchmarkdefaultSource
MATH Level 59.3%#84Epoch AI Benchmarking Hub

Knowledge

BenchmarkdefaultSource
LegalBench55.8%#129Vals AI
MedQA53.2%#85Vals AI

Human preference

BenchmarkdefaultSource
LMArena Text1197#288LMArena

Composite

BenchmarkdefaultSource
Epoch Capabilities Indexnot in index118.4#170Epoch AI Benchmarking Hub
AA Intelligence Index5.1#512Artificial Analysis

What a task costs

Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.

WorkloadTokens in / outCostWith cachingTime
Chat reply400 / 300$0.0004––
Summarise a 30-page report12,000 / 600$0.0058––
Code edit6,000 / 1,500$0.0038––
Agentic coding session60,000 / 4,000$0.030––
Structured extraction2,000 / 200$0.0010––

See also

Best under $2/MBest open weightsAll Mistral AI models

Data as of 9 Sept 2026. Compare with another model.