Every model, and what it costs
Ten vendors behind one OpenAI-compatible endpoint, reachable from an Iranian network without a VPN. Prices are live Toman rates read from the same table the biller uses.
Models by vendor
Rates are per 1,000 tokens in Toman. A blank price means the model is in the catalogue but has no priced row yet, so it is not callable — that is a state worth seeing rather than one worth hiding.OpenAI
| Model | Type | Context | In / 1K | Out / 1K |
|---|---|---|---|---|
| gpt-5.5GPT-5 | Reasoning | 400K | 1,561 | 9,367 |
| gpt-5.2GPT-5 | Reasoning | 400K | 570 | 4,561 |
| gpt-5GPT-5 | Reasoning | 200K | 407 | 3,258 |
| gpt-5-miniGPT-5 | Reasoning | 200K | 81 | 652 |
| gpt-5-nanoGPT-5 | Chat | 200K | 18 | 141 |
| gpt-4.1GPT-4.1 | Chat | 1,000K | 652 | 2,606 |
| gpt-4.1-miniGPT-4.1 | Chat | 1,000K | 130 | 521 |
| gpt-4oGPT-4o | Chat | 128K | 815 | 3,258 |
| gpt-4o-miniGPT-4o | Chat | 128K | 53 | 212 |
| o1o-series | Reasoning | 200K | 4,683 | 18,734 |
| o3o-series | Reasoning | 200K | 652 | 2,606 |
| o3-minio-series | Reasoning | 128K | 358 | 1,434 |
| o4-minio-series | Reasoning | 200K | 358 | 1,434 |
| gpt-4-turboGPT-4 | Chat | 128K | 3,258 | 9,774 |
| gpt-3.5-turboGPT-3.5 | Chat | 16K | 163 | 489 |
| text-embedding-3-largeEmbeddings | Embedding | 8.192K | 42 | — |
| text-embedding-3-smallEmbeddings | Embedding | 8.192K | 6.52 | — |
Anthropic
| Model | Type | Context | In / 1K | Out / 1K |
|---|---|---|---|---|
| claude-fable-5Claude Fable | Reasoning | 200K | 3,122 | 15,611 |
| claude-opus-5Claude Opus | Reasoning | 200K | 1,561 | 7,806 |
| claude-opus-4-8Claude Opus | Reasoning | 200K | 1,561 | 7,806 |
| claude-sonnet-5Claude Sonnet | Reasoning | 200K | 652 | 3,258 |
| claude-opus-4-7Claude Opus | Reasoning | 200K | 1,561 | 7,806 |
| claude-sonnet-4-6Claude Sonnet | Chat | 200K | 977 | 4,887 |
| claude-haiku-4-5Claude Haiku | Chat | 200K | 326 | 1,629 |
| claude-3-5-sonnetClaude 3.5 | Chat | 200K | 977 | 4,887 |
| claude-3-5-haikuClaude 3.5 | Chat | 200K | 326 | 1,629 |
| Model | Type | Context | In / 1K | Out / 1K |
|---|---|---|---|---|
| gemini-3.1-pro-previewGemini 3 | Reasoning | 1,000K | 624 | 3,747 |
| gemini-3.5-flashGemini 3 | Chat | 1,000K | 489 | 2,932 |
| gemini-3.1-flash-liteGemini 3 | Chat | 1,000K | 81 | 489 |
| gemini-2.5-proGemini 2.5 | Reasoning | 1,000K | 390 | 3,122 |
| gemini-2.5-flashGemini 2.5 | Chat | 1,000K | 106 | 882 |
| gemini-2.5-flash-liteGemini 2.5 | Chat | 1,000K | 35 | 141 |
DeepSeek
| Model | Type | Context | In / 1K | Out / 1K |
|---|---|---|---|---|
| deepseek-chatDeepSeek | Chat | 64K | 98 | 391 |
| deepseek-reasonerDeepSeek | Reasoning | 64K | 98 | 391 |
Priced, not yet documented
Callable today; the catalogue entry has not caught up. The id is what you put in the request body.- gpt-6.1-sol652 T / 1K in
- gpt-6.1-terra326 T / 1K in
- gpt-6-astra3,122 T / 1K in
- gpt-6-luna35 T / 1K in
- gpt-6-sol652 T / 1K in
- gpt-5.61,303 T / 1K in
- gpt-5.6-luna71 T / 1K in
- gpt-5.6-sol1,303 T / 1K in
- gpt-5.6-terra652 T / 1K in
- gpt-5.5-pro9,367 T / 1K in
- gpt-5.4815 T / 1K in
- gpt-5.4-mini244 T / 1K in
- gpt-5.4-nano65 T / 1K in
- gpt-5.4-pro9,367 T / 1K in
- gpt-5.3-chat-latest570 T / 1K in
- gpt-5.3-codex570 T / 1K in
- gpt-5.2-chat-latest570 T / 1K in
- gpt-5.2-codex570 T / 1K in
- gpt-5.2-pro6,557 T / 1K in
- gpt-5.1407 T / 1K in
- gpt-5.1-chat-latest407 T / 1K in
- gpt-5.1-codex407 T / 1K in
- gpt-5.1-codex-max407 T / 1K in
- gpt-5.1-codex-mini81 T / 1K in
- gpt-5-chat407 T / 1K in
- gpt-5-chat-latest407 T / 1K in
- gpt-5-codex407 T / 1K in
- gpt-5-pro4,683 T / 1K in
- gpt-5-search-api407 T / 1K in
- gpt-4-06139,774 T / 1K in
- gpt-49,774 T / 1K in
- o1-pro46,834 T / 1K in
- chat-latest1,629 T / 1K in
- claude-haiku-5-533 T / 1K in
- claude-opus-5-51,249 T / 1K in
- claude-sonnet-5-5652 T / 1K in
- claude-fable-5-13,122 T / 1K in
- claude-opus-4-61,561 T / 1K in
- claude-opus-4-51,561 T / 1K in
- claude-sonnet-4-5977 T / 1K in
- deep-research-pro-preview-12-2025652 T / 1K in
- gemma-4-31b-it33 T / 1K in
- gemma-4-26b-a4b-it33 T / 1K in
- deep-research-max-preview-04-2026652 T / 1K in
- deep-research-preview-04-2026652 T / 1K in
- gemini-4-argon652 T / 1K in
- gemini-3.8-flash244 T / 1K in
- gemini-3.8-live244 T / 1K in
- gemini-3.7-flash244 T / 1K in
- gemini-3.6-flash244 T / 1K in
- gemini-3.5-flash-lite98 T / 1K in
- gemini-3.1-flash-lite-preview81 T / 1K in
- gemini-3.1-flash-live-preview244 T / 1K in
- gemini-3.1-pro-preview-customtools624 T / 1K in
- gemini-3-flash-preview163 T / 1K in
- gemini-embedding-265 T / 1K in
- gemini-embedding-2-preview65 T / 1K in
- gemini-embedding-00149 T / 1K in
- gemini-flash-latest244 T / 1K in
- gemini-flash-lite-latest98 T / 1K in
- gemini-pro-latest652 T / 1K in
- deepseek-v4-flash106 T / 1K in
- deepseek-v4-flash-vision-exp98 T / 1K in
- deepseek-v4-pro430 T / 1K in
- deepseek-v3.298 T / 1K in
- deepseek-v398 T / 1K in
- deepseek-r198 T / 1K in
- deepseek-coder98 T / 1K in
- deepseek-flash98 T / 1K in
- grok-4.6652 T / 1K in
- kimi-k31,024 T / 1K in
- kimi-k2.7-code333 T / 1K in
- kimi-k2.6333 T / 1K in
- glm-5.3456 T / 1K in
- glm-5.3-flash49 T / 1K in
- glm-5.3-flashx121 T / 1K in
- glm-5.2456 T / 1K in
- glm-5.1456 T / 1K in
- qwen3.8-flash49 T / 1K in
- qwen3.8-max652 T / 1K in
- qwen3.7-max815 T / 1K in
- qwen3.6-flash81 T / 1K in
- qwen3.6-plus163 T / 1K in
- MiniMax-M2.7108 T / 1K in
- MiniMax-M2.7-highspeed215 T / 1K in
- hy4-preview307 T / 1K in
- hy351 T / 1K in
One endpoint, one key
- Base URL
https://1xai.ir/v1- Auth
Authorization: Bearer 1xai-…- Model selection
- The `model` field in the request body
- Streaming
- Supported on every chat and reasoning model
- Billing
- Per token, in Toman, from your wallet
The full route list, error shapes and the Anthropic-native passthrough are on the endpoint page.Per-model write-ups
Each model has a longer analysis — what it is good at, where it falls down, and which one to reach for instead. Those are written in Persian and live at 1xai.ir/models (Persian). The structured facts above are the same in both languages.Do I need a different key or base URL per provider?
No. One key and one base URL reach all ten. You select the vendor by the model name in the request body — 1xAi resolves gpt-*, claude-*, gemini-*, deepseek-*, grok-*, kimi-*, glm-*, qwen*, minimax-* and hy* to the right upstream. That is the whole point of an OpenAI-compatible endpoint: your existing client library keeps working and only the model string changes.
Are these prices final?
They are the customer-facing Toman rate, which is the upstream list price plus twenty percent, converted at the current FX rate. This table is rendered from the same table the biller reads, so what you see here is what a call costs. It moves when the FX rate moves.
What happens when a model is retired upstream?
It disappears from this table, because the table is the live catalogue rather than a document someone remembers to edit. Dated snapshot ids (gpt-4o-2024-08-06) are priced as their family, so pinning a snapshot never produces an unpriced call.
Is there an English description of each model?
Not yet — the per-model write-ups are Persian. The structured facts here (context window, endpoints, streaming, family) are the same in either language, and the vendors' own model cards cover the rest better than a translation would.
- API documentation
Quickstart, auth, native Anthropic/Gemini routes, streaming, Toman billing and account endpoints.
- Terms of Service
Billing, refunds, availability, liability and termination.
- Privacy Policy
What we collect, what never leaves, and what the model provider receives.
- Contact
Business enquiries, large top-ups, and anything the docs did not answer.
- Overview
What 1xAi is, who it is for, and what it costs — the English entry point.
- ChatGPT API from Iran
Whether the OpenAI API can be called from an Iranian network, and how.
- Claude API from Iran
Anthropic's Messages API from Iran, including the native passthrough route.
- OpenAI-compatible endpoint
The base URL, the auth header, every supported route, streaming and errors.
- Pricing
How billing works: list price plus a 15–30% markup by model, paid in Toman.
- Service status
Live per-provider reachability measured from inside Iran, plus the raw dataset.
- AI tutor for the Konkur
The learning panel for Iran's university entrance exam, compared with Zerang, Andarz, ChatGPT and Gemini, with sources.