GLM-5.3-Flash vs Llama-4-Scout-17B-16E-Instruct
Specs, VRAM requirements and download trends — updated 24 October 2026.
Specification comparison
Differences are highlighted; identical values are muted.
| Specification | GLM-5.3-Flash | Llama-4-Scout-17B-16E-Instruct |
|---|---|---|
| Deployment | Open weights · also via API | Open weights · also via API |
| Parameters | 321.3 B | 108.6 B |
| Architecture | Glm5NextForConditionalGeneration | Llama4ForConditionalGeneration |
| Context length | 1,048,575 | 327,680 |
| Licence | mit | other |
| Commercial use | Allowed | Unknown |
| Languages | 2 | 12 |
| Downloads 30d | 6,319,699 | 144,941 |
| Downloads all time | 10.3 M | 7.8 M |
| Quantizations on the Hub | SAFETENSORS | SAFETENSORS |
| Gated | No | Yes — accept licence |
| First seen on the Hub | 2026-08-25 | 2025-04-02 |
| API input price / 1M tokens | $0.15 | $0.10 |
| API output price / 1M tokens | $0.50 | $0.30 |
| Est. cost per 1M input + 1M output tokens | $0.65 | $0.40 |
| Cached input / 1M tokens | $0.03 | — |
| AA Intelligence Index | 41.8 | 6.5 |
| AA Coding Index | 71.5 | 8.2 |
| AA Agentic Index | 50.9 | 0.5 |
| Max output tokens | 943,717 | 16,384 |
| API providers | 34 | 2 |
| Tool calling | Yes | Yes |
| Reasoning mode | Yes | No |
API cost
Estimated cost of 1M input + 1M output tokens via OpenRouter
GLM-5.3-Flash
$0.65
Llama-4-Scout-17B-16E-Instruct
$0.40
Llama-4-Scout-17B-16E-Instruct is 1.6× cheaper at list price.
Download trend
Daily snapshots, last 56 days (30 Aug – 24 Oct)
6.6 M148 K
GLM-5.3-Flash
Llama-4-Scout-17B-16E-Instruct
VRAM side by side
On RTX 4090 · 24 GB · 8K context unless noted
Llama-4-Scout-17B-16E-Instruct · fp16
255.8 GB / 24 GB
GLM-5.3-Flash · fp16
409.9 GB / 24 GB