Model Comparison

Kimi K2-Thinking-0905 vs Mistral Large 3 (675B Instruct 2512)Which is better in 2026?

Kimi K2-Thinking-0905 significantly outperforms across most benchmarks. Mistral Large 3 (675B Instruct 2512) is 1.1x cheaper per token.

Verdict: Kimi K2-Thinking-0905 vs Mistral Large 3 (675B Instruct 2512) — which is better?

Kimi K2-Thinking-0905 (by Moonshot AI) and Mistral Large 3 (675B Instruct 2512) (by Mistral AI) are two of the AI models people compare most. Here is how they stack up on benchmarks, price and capabilities, and which one to pick in 2026.

Kimi K2-Thinking-0905 outperforms in 1 benchmarks (GPQA), while Mistral Large 3 (675B Instruct 2512) is better at 0 benchmarks. Kimi K2-Thinking-0905 significantly outperforms across most benchmarks.

On price, Mistral Large 3 (675B Instruct 2512) is roughly 1.1x cheaper per token on a blended 3:1 input/output basis, which adds up quickly at production volume.

Kimi K2-Thinking-0905 also accepts a larger context window (262,144 input tokens), making it the stronger choice for long documents and large codebases.

Choose Kimi K2-Thinking-0905 if…

  • you want the strongest raw capability — it leads on 1 of 1 shared benchmarks
  • you process long inputs — it offers a 262,144 token context window

Choose Mistral Large 3 (675B Instruct 2512) if…

  • cost matters — it's about 1.1x cheaper per token
  • you want the most recent training data — it shipped Dec 2025
Want to compare interactively?Try the playground

Performance Benchmarks

Comparative analysis across standard metrics

1 benchmarks

Kimi K2-Thinking-0905 outperforms in 1 benchmarks (GPQA), while Mistral Large 3 (675B Instruct 2512) is better at 0 benchmarks.

Kimi K2-Thinking-0905 significantly outperforms across most benchmarks.

Sun Jun 07 2026 • llm-stats.com

Arena Performance

Human preference votes

Pricing Analysis

Price comparison per million tokens

Mistral Large 3 (675B Instruct 2512) costs less

For input processing, Kimi K2-Thinking-0905 ($0.47/1M tokens) is 1.1x cheaper than Mistral Large 3 (675B Instruct 2512) ($0.50/1M tokens).

For output processing, Kimi K2-Thinking-0905 ($2.00/1M tokens) is 1.3x more expensive than Mistral Large 3 (675B Instruct 2512) ($1.50/1M tokens).

In conclusion, Kimi K2-Thinking-0905 is more expensive than Mistral Large 3 (675B Instruct 2512).*

* Using a 3:1 ratio of input to output tokens

Lowest available price from all providers
Sun Jun 07 2026 • llm-stats.com
Moonshot AI
Kimi K2-Thinking-0905
Input tokens$0.47
Output tokens$2.00
Best providerDeepinfra
Mistral AI
Mistral Large 3 (675B Instruct 2512)
Input tokens$0.50
Output tokens$1.50
Best providerMistral
Notice missing or incorrect data?Start an Issue→

Model Size

Parameter count comparison

325.0B diff

Kimi K2-Thinking-0905 has 325.0B more parameters than Mistral Large 3 (675B Instruct 2512), making it 48.1% larger.

Moonshot AI
Kimi K2-Thinking-0905
1.0Tparameters
Mistral AI
Mistral Large 3 (675B Instruct 2512)
675.0Bparameters
1000.0B
Kimi K2-Thinking-0905
675.0B
Mistral Large 3 (675B Instruct 2512)

Context Window

Maximum input and output token capacity

Kimi K2-Thinking-0905 accepts 262,144 input tokens compared to Mistral Large 3 (675B Instruct 2512)'s 262,100 tokens. Kimi K2-Thinking-0905 can generate longer responses up to 262,144 tokens, while Mistral Large 3 (675B Instruct 2512) is limited to 262,100 tokens.

Moonshot AI
Kimi K2-Thinking-0905
Input262,144 tokens
Output262,144 tokens
Mistral AI
Mistral Large 3 (675B Instruct 2512)
Input262,100 tokens
Output262,100 tokens
Sun Jun 07 2026 • llm-stats.com

Input Capabilities

Supported data types and modalities

Mistral Large 3 (675B Instruct 2512) supports multimodal inputs, whereas Kimi K2-Thinking-0905 does not.

Mistral Large 3 (675B Instruct 2512) can handle both text and other forms of data like images, making it suitable for multimodal applications.

Kimi K2-Thinking-0905

Text
Images
Audio
Video

Mistral Large 3 (675B Instruct 2512)

Text
Images
Audio
Video

License

Usage and distribution terms

Kimi K2-Thinking-0905 is licensed under MIT, while Mistral Large 3 (675B Instruct 2512) uses Apache 2.0.

License differences may affect how you can use these models in commercial or open-source projects.

Kimi K2-Thinking-0905

MIT

Open weights

Mistral Large 3 (675B Instruct 2512)

Apache 2.0

Open weights

Release Timeline

When each model was launched

Kimi K2-Thinking-0905 was released on 2025-09-05, while Mistral Large 3 (675B Instruct 2512) was released on 2025-12-04.

Mistral Large 3 (675B Instruct 2512) is 3 months newer than Kimi K2-Thinking-0905.

Kimi K2-Thinking-0905

Sep 5, 2025

9 months ago

Mistral Large 3 (675B Instruct 2512)

Dec 4, 2025

6 months ago

3mo newer

Knowledge Cutoff

When training data ends

Neither model specifies a knowledge cutoff date.

Unable to compare the recency of their training data.

No cutoff dates available

Provider Availability

Kimi K2-Thinking-0905 is available from DeepInfra, Novita, Fireworks. Mistral Large 3 (675B Instruct 2512) is available from Mistral AI.

Kimi K2-Thinking-0905

deepinfra logo
Deepinfra
Input Price:Input: $0.47/1MOutput Price:Output: $2.00/1M
novita logo
Novita
Input Price:Input: $0.48/1MOutput Price:Output: $2.00/1M
fireworks logo
Fireworks
Input Price:Input: $0.60/1MOutput Price:Output: $2.50/1M

Mistral Large 3 (675B Instruct 2512)

mistral logo
Mistral
Input Price:Input: $0.50/1MOutput Price:Output: $1.50/1M
* Prices shown are per million tokens

Outputs Comparison

Notice missing or incorrect data?Start an Issue discussion→

Key Takeaways

Moonshot AI
Larger context window (262,144 tokens)
Less expensive input tokens
Higher GPQA score (84.5% vs 43.9%)
Mistral AI
Supports multimodal inputs
Less expensive output tokens

Detailed Comparison

AI Model Comparison Table
Feature
Moonshot AI
Kimi K2-Thinking-0905
Mistral AI
Mistral Large 3 (675B Instruct 2512)

FAQ

Common questions about Kimi K2-Thinking-0905 vs Mistral Large 3 (675B Instruct 2512).

Which is better, Kimi K2-Thinking-0905 or Mistral Large 3 (675B Instruct 2512)?

Kimi K2-Thinking-0905 significantly outperforms across most benchmarks. Kimi K2-Thinking-0905 is made by Moonshot AI and Mistral Large 3 (675B Instruct 2512) is made by Mistral AI. The best choice depends on your use case — compare their benchmark scores, pricing, and capabilities above.

How does Kimi K2-Thinking-0905 compare to Mistral Large 3 (675B Instruct 2512) in benchmarks?

Kimi K2-Thinking-0905 scores AIME 2025: 100.0%, HMMT 2025: 97.5%, MMLU-Redux: 94.4%, FRAMES: 87.0%, MMLU-Pro: 84.6%. Mistral Large 3 (675B Instruct 2512) scores MMMLU: 85.5%, AMC_2022_23: 52.0%, GPQA: 43.9%, LiveCodeBench: 34.4%, SimpleQA: 23.8%.

Is Kimi K2-Thinking-0905 cheaper than Mistral Large 3 (675B Instruct 2512)?

Kimi K2-Thinking-0905 is 1.1x cheaper for input tokens. Kimi K2-Thinking-0905 costs $0.47/M input and $2.00/M output via deepinfra. Mistral Large 3 (675B Instruct 2512) costs $0.50/M input and $1.50/M output via mistral.

What are the context window sizes for Kimi K2-Thinking-0905 and Mistral Large 3 (675B Instruct 2512)?

Kimi K2-Thinking-0905 supports 262K tokens and Mistral Large 3 (675B Instruct 2512) supports 262K tokens. A larger context window lets you process longer documents, conversations, or codebases in a single request.

What are the main differences between Kimi K2-Thinking-0905 and Mistral Large 3 (675B Instruct 2512)?

Key differences include context window (262K vs 262K), input pricing ($0.47 vs $0.50/M), multimodal support (no vs yes), licensing (MIT vs Apache 2.0). See the full comparison above for benchmark-by-benchmark results.

Who makes Kimi K2-Thinking-0905 and Mistral Large 3 (675B Instruct 2512)?

Kimi K2-Thinking-0905 is developed by Moonshot AI and Mistral Large 3 (675B Instruct 2512) is developed by Mistral AI.

Related comparisons