Gemini 3.5 Flash-Lite vs Kimi K3: Specs & Benchmark Comparison
Gemini 3.5 Flash-Lite is developed by Google, while Kimi K3 comes from Moonshot AI. Both were released in July 2026. Kimi K3 has a published size of about 2.8 trillion parameters; Google has not disclosed the parameter count of Gemini 3.5 Flash-Lite.
The two models share 2 published benchmarks. Kimi K3 leads on 2 of them. The widest gaps are on Terminal-Bench 2.1, where Kimi K3 scores 88.3% against 54.0%; CharXiv-R, where Kimi K3 scores 91.3% against 76.5%. Averaged across everything we track, Gemini 3.5 Flash-Lite sits at 53.2% and Kimi K3 at 67.8%.
Gemini 3.5 Flash-Lite is the cheaper API at $0.30 per million input tokens and $2.50 per million output tokens, roughly 10 times cheaper than Kimi K3 at $3 and $15. Both accept a context window of about 1M tokens. Both accept text and images as input.
| Characteristic | Gemini 3.5 Flash-Lite | Kimi K3 |
|---|---|---|
| Company | Moonshot AI | |
| Release Date | July 21, 2026 | July 16, 2026 |
| Parameters | — | 2.8T |
| Multimodal | Yes | Yes |
| Context (input) | 1.0M | 1.0M |
| Context (output) | 66K | 1.0M |
| Input Price / 1M | $0.30 | $3.00 |
| Output Price / 1M | $2.50 | $15.00 |
| Average Score | 53.2% | 67.8% |
| Benchmarks | ||
| Terminal-Bench 2.1 | 54.0% | 88.3% |
| CharXiv-R | 76.5% | 91.3% |
Visual Benchmark Comparison
Verdict
Gemini 3.5 Flash-Lite leads in 2 out of 4 comparison categories.
Both models show comparable average scores: Gemini 3.5 Flash-Lite — 0.5, Kimi K3 — 0.7.
Gemini 3.5 Flash-Lite is 6.4x cheaper: input $0.30/1M vs $3.00/1M tokens.
Gemini 3.5 Flash-Lite supports a larger context: 1M vs 1M tokens.
Both models were released around the same time: 7/21/2026 and 7/16/2026.
More About These Models
Related Comparisons
Frequently Asked Questions
Which is better for coding — Gemini 3.5 Flash-Lite or Kimi K3?
Which model is cheaper — Gemini 3.5 Flash-Lite or Kimi K3?
Which has a larger context window — Gemini 3.5 Flash-Lite or Kimi K3?
The Gemini 3.5 Flash-Lite and Kimi K3 comparison is updated for 2026. Data includes benchmark results, API pricing, context window size and other specifications. For more detailed information, visit the Gemini 3.5 Flash-Lite or Kimi K3 page. See also the complete list of AI model comparisons.