Compare / GLM-4.7-Flash vs Kanana-2 30B-A3B Thinking
GLM-4.7-Flash vs Kanana-2 30B-A3B Thinking
Updated November 2026
Verdict
GLM-4.7-Flash scores higher on the Small LM Index: 66.1 vs 50.7. On the 3 benchmarks both labs report, GLM-4.7-Flash wins 3 and Kanana-2 30B-A3B Thinking wins 0. Both need about 20.4 GB of memory at 4-bit quantization. Only GLM-4.7-Flash has a hosted API on OpenRouter ($0.15 per million tokens); the other has to be self-hosted.
Pick GLM-4.7-Flash if…
- you want the strongest model overall
- you mostly write code
- you need math and reasoning
- you want a hosted API rather than running it yourself
- you work with long documents
- you need a permissive license (MIT)
- you want the more widely used model (more quantized builds and community support)
Pick Kanana-2 30B-A3B Thinking if…
No clear advantage in the data we have.
Specifications
| GLM-4.7-Flash | Kanana-2 30B-A3B Thinking | |
|---|---|---|
| Small LM Index | 66.1 (likely 63–70) | 50.7 (likely 46–55) |
| Lab | Z.ai (GLM) · China | Kakao (Kanana) · Rest of Asia |
| Released | 2026-01-19 | 2026-01-14 |
| Parameters | 31B (3B active, MoE) | 30.7B (3B active, MoE) |
| Memory at 4-bit | ~20.4 GB | ~20.2 GB |
| Context window | 200K tokens | 32K tokens |
| Input | text | text |
| Reasoning | Always thinks | Always thinks |
| License | MIT | Kanana License |
| API price / 1M tokens | $0.15 ($0.061 in, $0.40 out) | Self-host only |
| Hugging Face downloads (30 days) | 956.4K | 1.3K |
Benchmarks
Higher is better. Bold marks the winner where both labs report a score. ≈ marks the index model's estimate for a score that wasn't reported, ✓ an independent measurement and * a score measured by another lab.
| Benchmark | GLM-4.7-Flash | Kanana-2 30B-A3B Thinking |
|---|---|---|
| MMLU-Pro Knowledge | ≈ 82 | 74.2 |
| GPQA Diamond Knowledge | 75.2 | 57.8 |
| Humanity's Last Exam Reasoning | 14.4 | ≈ 7 |
| AIME Math | 91.6 | 74.0 |
| LiveCodeBench Coding | 64.0 | 58.8 |
| SWE-bench Verified Coding | 59.2 | ≈ 32 |
| τ²-Bench Agentic | 79.5 | — |
| IFBench Instruction following | — | 47.8 |
| IFEval Instruction following | — | 82.2 |
Scores are mostly reported by each lab with its own settings, so small gaps are noise. See the methodology. Open in the interactive comparison to add more models.
Questions
Which is better, GLM-4.7-Flash or Kanana-2 30B-A3B Thinking?
GLM-4.7-Flash scores higher on the Small LM Index: 66.1 vs 50.7. On the 3 benchmarks both labs report, GLM-4.7-Flash wins 3 and Kanana-2 30B-A3B Thinking wins 0.
Which is cheaper to run?
Only GLM-4.7-Flash has a hosted API on OpenRouter ($0.15 per million tokens); the other has to be self-hosted. Both need about 20.4 GB of memory at 4-bit quantization.
Which has the longer context window?
GLM-4.7-Flash has the longer context window: 200K tokens against 32K.
Can I use GLM-4.7-Flash and Kanana-2 30B-A3B Thinking commercially?
GLM-4.7-Flash is released under MIT and Kanana-2 30B-A3B Thinking under Kanana License. Apache 2.0 and MIT allow commercial use; other licenses can add conditions, so read the license before shipping a product.