February 25: model added:gpt-5.3-codex,qwen3.5-flash,qwen3.5-122b-a10b,qwen3.5-27b,qwen3.5-35b-a3b

February 21: model added:qwen3-coder-next

February 20: model added:jina-embeddings-v5-text-nano,jina-embeddings-v5-text-small,gemini-3.1-pro-preview-customtools

February 19: model added:gemini-3.1-pro-preview

February 18: model added:claude-sonnet-4-6,claude-sonnet-4-6-think

February 16: model added:qwen3.5-plus,qwen3.5-397b-a17b

February 14: model added:doubao-seed-2-0-pro,doubao-seed-2-0-code-preview,doubao-seed-2-0-lite,doubao-seed-2-0-mini

February 8: Model deprecation notice — OpenAI will retire chatgpt-4o-latest on Feb 17, 2026. Requests will be automatically routed to gpt-4o-2024-11-20 after the shutdown

February 6: model added:claude-opus-4-6,claude-opus-4-6-think

February 2: model added: step-3.5-flash, step-3.5-flash-free

January 27:model added:kimi-k2.5,qwen3-max-2026-01-23

January 19:model added:glm-4.7-flash-free

January 13:model added:gemini-2.5-flash-preview-tts,gemini-2.5-pro-preview-tts

December 23:model added:mimo-v2-flash,cc-glm-4.7,coding-glm-4.7,coding-glm-4.7-free,crush-glm-4.7-free,minimax-m2.1,coding-minimax-m2.1,coding-minimax-m2.1-free

December 18: model added:mimo-v2-flash-free,doubao-seed-1-8

December 17: model added:gpt-image-1.5,gemini-3-flash-preview

December 15: New Feature: Google API Supports Files API

December 12: model added:gpt-5.2,gpt-5.2-pro,gpt-5.2-chat-latest

December 9: model added: deepseek-v3.2-fast,glm-4.6v

December 2: model added: mistral-large-3,deepseek-math-v2,doubao-seedream-4-5

December 2: model added: deepseek-v3.2, deepseek-v3.2-think

December 1: model added: deepseek-v3.2-speciale

November 27:model added: jimeng-3.0-pro, jimeng-3.0-720p, jimeng-3.0-1080p

November 25: The login system was upgraded to support multiple sign-in methods, For any issues, please contact us.feedback@aihubmix.com

November 25:model added: claude-opus-4-5-20251101, with pricing consistent with the official rates.

November 20:models added:gemini-3-pro-image-preview

November 19:models added:gemini-3-pro-preview,gemini-3-pro-preview-search

November 17: Added a new video model: veo-3.1-fast-generate-preview

November 16:Added a new video model: wan2.5-i2v-preview, wan2.5-t2v-preview, wan2.2-i2v-plus, wan2.2-t2v-plus

November 15:models added:gpt-5.1, gpt-5.1-codex, gpt-5.1-codex-mini, gpt-5.1-chat-latest

November 13:models added:ernie-5.0-thinking-preview

November 12:models added:kimi-k2-thinking

November 4:models added:qwen3-max-preview (official promotional name Qwen3-Max-Thinking), doubao-seed-1-6-251015, doubao-seed-1-6-lite-251015, doubao-seed-1-6-flash-250828, doubao-seed-1-6-thinking-250715.

November 3rd: Added a new video model: veo-3.1-generate-preview

October 26:New models added:deepseek-ai/DeepSeek-OCR

October 24:New models added:inclusionAI/Ling-1T, inclusionAI/Ring-1T, inclusionAI/Ring-flash-2.0

October 19:1. Added official API support for OpenAI’s video generation models sora-2 and sora-2-pro; see request examples at https://aihubmix.com/model/sora-2  

2. The Qwen VL model now supports video input; see request examples at https://aihubmix.com/model/qwen3-vl-30b-a3b-instruct  

3. The Doubao 1.6 series and Zhipu 4.6 series models now support a parameter switch for chain-of-thought mode; see request examples at https://aihubmix.com/model/glm-4.6

October 15:New models added:claude-haiku-4-5

October 11:New models added:embedding-3,embedding-2,gpt-image-1-mini

October 10:New models added:qwen3-vl-30b-a3b-thinking,qwen3-vl-30b-a3b-instruct

October 9: Added new reranking model jina-reranker-v3.

September 30:New models added:claude-sonnet-4-5-20250929,glm-4.6

September 29:New models added: DeepSeek-V3.2-Exp and DeepSeek-V3.2-Exp-Think.

September 28:Newly added Baidu series models: ERNIE-4.5-21B-A3B-Thinking, ERNIE-4.5-Turbo-VL-Latest, ERNIE-4.5-Turbo-Latest, ERNIE-4.5-0.3B, ERNIE-4.5-Turbo-VL.

September 25:DeepSeek-V3.1 and DeepSeek-V3.1-Think were both upgraded to the latest version of Terminus.

September 24: New models added: gpt-5-codex, qwen3-max, qwen3-vl-plus, qwen3-vl-235b-a22b-thinking, qwen3-vl-235b-a22b-instruct

September 23: The gemini-1.5-pro and gemini-1.5-flash models will be officially phased out soon.

September 22 Qwen series qwen-image-edit, Doubao seedream4, and Baidu series drawing models

September 6: Added new model qwen3-max-preview

August 27: gemini-2.5-flash-image-preview

August 19: deepseek-chat upgraded to DeepSeek-V3.1

August 11：new model：glm-4.5v

August 8: New additions: gpt-5、gpt-5-mini, gpt-5-nano, gpt-5-chat-latest; these three models are now supported;

August 6：New model claude-opus-4-1-20250805、openai/gpt-oss-20b、openai/gpt-oss-120b、qwen-flash、qwen3-coder-flash

July 23: New model qwen3-coder-480b-a35b-instruct、gemini-2.5-flash-lite

July 23: New model Qwen/Qwen3-235B-A22B-Instruct-2507

July 12: New model kimi-k2-0711-preview added

July 10: New model grok-4 added

June 27: Added models o4-mini-deep-research and o3-deep-research; only supported through v1/responses interface requests, with necessary parameters including 'web_search_preview' or 'mcp' tools.

June 20: o3-pro price decreased by 10% and is now consistent with the OpenAI official website price.

June 18: New models gemini-2.5-flash and gemini-2.5-pro gemini-2.5-flash-lite-preview-06-17added.

June 14: O3 price decreased by 80%, consistent with OpenAI official.

June 13: Added Google’s latest text-to-video model veo-3.0-generate-preview.

June 11: Added model o3-pro; we will definitely adjust the o3 price to be consistent with OpenAI's official pricing, but we need to wait for Microsoft's notification. Please be patient.

Due to the increased maintenance costs of the reverse interface, the price per call has been raised. (Update date: 2025-06-04)；

gpt-4o-image-vip: After adjustment $0.009 per use, before adjustment $0.007 per use  

gpt-4o-image: After adjustment $0.005 per use, before adjustment $0.003 per use

May 30：aihubmix's DeepSeek-R1 has been automatically upgraded to the latest version 250528; inference thinking data and result data are sent separately.

May 29：deepseek-ai/DeepSeek-R1-0528

May 23: Added models claude-sonnet-4-20250514 and claude-opus-4-20250514、gemini-2.5-flash-preview-05-20-search

May 22: Added new model aihubmix-router; requested the aihubmix-router to automatically route models based on question complexity, so everyone no longer needs to manually switch models;

Currently, there are four routing models: gpt-4.1, gpt-4.1-mini, gpt-4.1-nano, o4-mini

May 21: Added model gemini-2.5-flash-preview-05-20、imagen-4.0-generate-preview-05-20

May 14:

New models added: kimi-thinking-preview, nvidia/Llama-3_1-Nemotron-Ultra-253B-v1, gpt-4o-mini-tts

Special note: The billing for gpt-4o-mini-tts may be more expensive or cheaper than the official rates for some requests. Because the billing is very complex, we use an estimated method. We will strive to keep the discrepancy with official charges within 5% based on extensive research.

May 9: Added Ideogram's V3 text-to-image model API

May 6: Added models gemini-2.5-pro-preview-05-06 and gemini-2.5-pro-preview-05-06-search

April 30: Added deepseek-ai/DeepSeek-Prover-V2-671B

April 29: Added Qwen3 series models

Added the latest official image generation model; gpt-image-1, with prices consistent with the official rates; (note that besides the listed image generation charges and input text fees, there are additional official deductions) overall pricing is still relatively high; therefore, we will temporarily continue to retain the reverse interface to lower the usage threshold for customers trying out new features.

April 22: New models added: Doubao-1.5-thinking-pro and Aihubmix-MAI-DS-R1

April 19: Change the model name to gemini-2.5-flash-preview-04-17-nothink to disable the thinking mode of gemini-2.5-flash-preview-04-17-nothink.

April 17: Added new models o4-mini, o3

THUDM/GLM-Z1-32B-0414、THUDM/GLM-4-32B-0414、THUDM/GLM-Z1-9B-0414、THUDM/GLM-4-9B-0414

April 16: Aihubmix: Added support for OpenAI's Responses-API interface https://docs.aihubmix.com/en/api/Responses-API

April 15: New model added, OpenAI's 4.1 series model.

ntegrates Tavily search capabilities, allowing all models to connect to the internet; simply add :surfing after the model ID; https://docs.aihubmix.com/cn/api/LLM-Search

April 9: New Ideogram series text-to-image model added, with support for Claude caching functionality.

April 6th: Nen model Llama-4: chutesai/Llama-4-Maverick-17B-128E-Instruct-FP8, chutesai/Llama-4-Scout-17B-16E-Instruct.

New model: gemini-2.5-pro-exp-03-25、deepseek-ai/DeepSeek-V3-0324、unsloth/gemma-3-27b-it、unsloth/gemma-3-12b-it、chutesai/Mistral-Small-3.1-24B-Instruct-2503、Moonlight-16B-A3B

March 16: Added support for native search functionality for OpenAI and Google Gemini models; in the future, this interface will be improved to extend support for third-party searches.

March 15: New models added: gpt-4o-mini-search-preview and gpt-4o-search-preview.