GLM-5.1 and Kimi K2.7 on one OpenAI-compatible API — for any app that needs an LLM. Billed in Egyptian pounds.
We benchmark on coding because it's the hardest public test of reasoning — but these models power any LLM workload: chat, agents, extraction, RAG.
Frontier-class coding at a fifth of the price, with no international card. Coding scores are independently-verified SWE-Bench Verified results for the last measured releases (GLM-5, Kimi K2.6); the GLM-5.1 and K2.7 served on Tarmeez are vendor-reported a notch higher, with GLM-5.1 topping SWE-Bench Pro at 58.4. Prices are output per 1M tokens at ≈50 EGP/USD.
Per 1M tokens. New accounts get EGP 100 free.
| Model | Input | Output |
|---|---|---|
| GLM-5.1 reasoning & code | EGP 100 | EGP 320 |
| Kimi K2.7 Code coding & agents | EGP 70 | EGP 290 |
| DeepSeek V4 Flash budget & fast | EGP 10 | EGP 20 |
| GLM-5.2 coding-first, 1M context | Coming soon | |
Note: GLM-5.2 opens when its open weights ship, around June 20.
Point any OpenAI SDK at Tarmeez.
Balance shown in EGP, always.
Switch models with one string. Pay per token.
Egyptian-hosted models are coming soon: an in-country tier under Egypt's data-protection law, built for banks and government that need their data to stay home.
Run GLM-5.1 and Kimi K2.7 on your own workloads.
Create your API key →