Codestral 25.08 vs Kimi K3

Codestral 25.08 and Kimi K3 are both current production-tier models. Codestral 25.08 is meaningfully cheaper at $0.30 / $0.90 per 1M. Kimi K3 has a 1.0M context window — about 4× the 256k of Codestral 25.08. Kimi K3 leads on coding, reasoning, general knowledge.

Specs side by side

Metric
Mistral AI
Codestral 25.08
Z.ai
Kimi K3
Input price (per 1M)$0.3$3
Output price (per 1M)$0.9$15
Context window256k tokens1.0M tokens
Speed tierfastbalanced
Open weightsYesNo
EU regionYesNo
Free tierLa PlateformeNo
Prompt cachingNoYes
Vision inputNoYes
Extended thinkingNoYes

When to choose each

Mistral AI Free tier

Choose Codestral 25.08 if…

  • Cost is a priority ($0.30 / $0.90 per 1M vs $3 / $15 per 1M)
  • You need open weights for self-hosting or fine-tuning
  • EU data residency is required
  • HIPAA eligibility is required
  • You want a free tier for prototyping
Z.ai

Choose Kimi K3 if…

  • You need 1.0M context (4× more than Codestral 25.08)
  • You need image input / vision
  • Coding is central to your workload
  • Reasoning is central to your workload

Benchmark delta

Codestral 25.08 leads on

Codestral 25.08 has no meaningful benchmark lead in this pair.

Kimi K3 leads on

  • Coding
  • Reasoning
  • General knowledge
  • Long-context retrieval
  • Instruction following
  • Multilingual
  • Tool use

FAQ — Codestral 25.08 vs Kimi K3

Codestral 25.08 vs Kimi K3 — which is better?

Codestral 25.08 and Kimi K3 are both current production-tier models. Codestral 25.08 is meaningfully cheaper at $0.30 / $0.90 per 1M. Kimi K3 has a 1.0M context window — about 4× the 256k of Codestral 25.08. Kimi K3 leads on coding, reasoning, general knowledge. The right pick depends on your use case — see "When to choose each" above for a data-driven decision.

How does Codestral 25.08 pricing compare to Kimi K3?

Codestral 25.08 costs $0.30 / $0.90 per 1M vs Kimi K3 at $3 / $15 per 1M. Codestral 25.08 is cheaper on output tokens by roughly 1567%. Both support prompt caching, which reduces effective cost by 80-90% on repeat system prompts.

Does Codestral 25.08 or Kimi K3 have the bigger context window?

Kimi K3 has a 1.0M-token context window — 4× the 256k context of Codestral 25.08. Enough for entire codebases, books, or multi-document RAG.

Is there a free tier for Codestral 25.08 or Kimi K3?

Codestral 25.08: yes — Free tier for prototyping with rate limits. Kimi K3: no — No free API tier; kimi.com web chat is free with limits.

Which is better for coding — Codestral 25.08 or Kimi K3?

Kimi K3 leads on coding benchmarks (Codestral 25.08: 90/100, Kimi K3: 95/100). For production coding agents also weigh tool-use performance — Codestral 25.08 scores 78, Kimi K3 scores 90.