Chat and text models, at today's Rial prices

A chat model takes text and writes text. It is the most used kind: answering, summarising, writing, coding and agents. The list below shows every active text model at today's Rial price.

Models and today's Rial prices

36 models available now. Prices are per million tokens, read straight from the catalogue.

ModelInput (IRR)Output (IRR)
DeepSeekDeepSeek: DeepSeek V4.1 Flashdeepseek/deepseek-v4.1-flash824K3.3M
Z.aiZ.ai: GLM 5.3 Flashz-ai/glm-5.3-flash110K215K
NVIDIANVIDIA: Nemotron 3 Ultranvidia/nemotron-3-ultra-550b-a55b1.37M6.04M
Z.aiZ.ai: GLM 5.3z-ai/glm-5.3192K3.84M
AnthropicAnthropic: Claude Haiku 5.5anthropic/claude-haiku-5.5275K1.37M
DeepSeekDeepSeek: DeepSeek V4 Flash 0731deepseek/deepseek-v4-flash-0731122K824K
GoogleGoogle: Gemini 3.8 Flashgoogle/gemini-3.8-flash2.06M10.3M
NVIDIANVIDIA: Nemotron 3.5 Lightningnvidia/nemotron-3.5-lightning192K549K
QwenQwen: Qwen3.8 Flashqwen/qwen3.8-flash412K1.29M
Z.aiZ.ai: GLM 5.2z-ai/glm-5.21.7M6.65M
Moonshot AIMoonshotAI: Kimi K3moonshotai/kimi-k3527K43.94M
DeepSeekDeepSeek: DeepSeek V4 Pro 0813deepseek/deepseek-v4-pro-08133.62M10.87M
GoogleGoogle: Gemma 4 26B A4Bgoogle/gemma-4-26b-a4b-it247K824K
QwenQwen: Qwen3.8 27Bqwen/qwen3.8-27b165K5.38M
MinimaxMiniMax: MiniMax M3minimax/minimax-m3824K3.3M
GoogleGoogle: Gemma 4 31Bgoogle/gemma-4-31b-it549K1.1M
AnthropicAnthropic: Claude Fable 5anthropic/claude-fable-527.46M137.3M
AnthropicAnthropic: Claude Fable 5.1anthropic/claude-fable-5.127.46M137.3M
AnthropicClaude Opus 5anthropic/claude-opus-513.73M68.65M
AnthropicAnthropic: Claude Sonnet 5anthropic/claude-sonnet-55.49M27.46M
GoogleGoogle: Gemini 3.5 Flashgoogle/gemini-3.5-flash4.12M24.71M
GoogleGoogle: Gemini 3.7 Flashgoogle/gemini-3.7-flash2.06M10.3M
Meituan: LongCat 2.0meituan/longcat-2.0824K3.3M
MetaMeta: Muse Glimmer 30Bmeta/muse-glimmer-30b961K4.12M
Moonshot AIMoonshotAI: Kimi K2.7 Codemoonshotai/kimi-k2.7-code1.65M8.24M
OpenAIOpenAI: GPT-5.6 Solopenai/gpt-5.6-sol10.98M54.92M
OpenAIOpenAI: GPT-6 Astraopenai/gpt-6-astra27.46M137.3M
PoolsidePoolside: Laguna S 2.1poolside/laguna-s-2.1247K494K
QwenQwen: Qwen3.7 Flashqwen/qwen3.7-flash82K357K
QwenQwen: Qwen3.7 Maxqwen/qwen3.7-max4.05M12.15M
QwenQwen: Qwen3.7 Plusqwen/qwen3.7-plus879K3.51M
QwenQwen: Qwen3.8 2.4T A95Bqwen/qwen3.8-2.4t-a95b5.49M16.48M
TencentTencent: Hy-MT2-30B-A3Btencent/hy-mt2-30b-a3b203K810K
TencentTencent: Hy3tencent/hy3227K906K
TencentTencent: Hy4 previewtencent/hy4-preview2.06M6.18M
xAISpaceXAI: Grok 4.7x-ai/grok-4.75.49M16.48M

How to choose

Price matters, but it is not the only measure. Work out what the job needs, then find the least costly model that can do it.

  • For long text or a large project, the context window has to hold your whole input.
  • If you send images or files, pick a model that accepts them.
  • Agents need tool calling, and reasoning quality matters a lot.

How the price works

Each model has two rates per million tokens: input for what you send and output for what you get back. An amount is held before each request and only the real cost is taken when the answer ends.

Connecting

All of them are called on one route: POST /v1/chat/completions. With code on the OpenAI SDK you change base_url and the key, and switching models means changing the model value.

Frequently asked questions

Which model is best for Persian?

No reliable index compares models in Persian. Give two or three of them a real text from your own work and judge the results.

Can I compare several models?

Yes. One ir- key reaches every active model, and the models page can put several side by side.

Do I pay for a failed request?

A failed request is not charged and the held amount is released. Every request is recorded in the dashboard.

Get connected today

Create an account, top up your wallet and make your first request in minutes — no contract, no foreign card.

Create a free accounthttps://iranrouter.com/v1