← All Comparisons

GLM-5 vs Mistral Medium 3

A detailed comparison of GLM-5 (Zhipu AI) and Mistral Medium 3 (Mistral) across pricing, performance, and features.

Pricing Comparison

MetricGLM-5Mistral Medium 3Difference
Input / 1M tokens$1.00$0.40-60%
Output / 1M tokens$3.20$2.00-38%
Context window200K128K
Max output128K16.384K

Benchmark Comparison

BenchmarkGLM-5Mistral Medium 3
MMLU-Pro70.4%76%
HumanEval91%87%
GPQA72%

Capabilities

CapabilityGLM-5Mistral Medium 3
code
reasoning
text
tool-use
vision

GLM-5 Strengths

  • Open-weight (MIT license) — self-hostable
  • 77.8% SWE-Bench Verified — top-tier coding
  • 128K max output — huge generation window

GLM-5 Weaknesses

  • MMLU-Pro lags behind Western flagships
  • 744B parameters — heavy to self-host
  • China-based — availability concerns

Mistral Medium 3 Strengths

  • Very affordable for mid-tier quality
  • Good multilingual performance

Mistral Medium 3 Weaknesses

  • Lags behind on reasoning benchmarks
  • Smaller community and tooling

Quick Verdict

Best value: Mistral Medium 3 is the more affordable option at $0.4/$2 per 1M tokens.

Higher benchmarks: Mistral Medium 3 scores higher on average across available benchmarks (81.5% avg).

Larger context: GLM-5 supports 200K tokens.

Choose Mistral Medium 3 if cost matters most. Choose GLM-5 if you need the best possible quality for complex tasks.

More Comparisons