📅 June 2026
⚠️ gpt-image-2-vip size is unavailable again; output is temporarily limited to 1K
June 23, 2026Temporary gpt-image-2-vip size parameter update
-
Based on customer reports, live testing, and code-level review this morning, default-group
gpt-image-2-viplost the restoredsizebehavior again after another upstream rule change -
gpt-image-2-vipcurrently generates 1K images normally; the submittedsizevalue is silently ignored, so 2K / 4K cannot be selected through this route for now -
The
gpt-image-2-vipmodel name and $0.03/call per-request billing remain unchanged -
For 2K / 4K or precise size control, switch to official-forward
gpt-image-2under theSora2OfficialorGPTImage2 Enterprisegroup -
Official-forward
gpt-image-2supports 4K / 2K and precisesizecontrol; it is billed by official input / output tokens, not by default-group per-call pricing - The June 18, 2026 restoration notice is superseded by this announcement
- 📝 Developer documentation:
🎨 gpt-image-2-vip Size and Quality Parameters Restored (Historical Notice)
June 18, 2026Historical gpt-image-2-vip parameter capability update
- This entry describes the short-lived restoration state on June 18, 2026; it is no longer current
-
At that time, default-group
gpt-image-2-viptemporarily restoredsizesupport -
Common explicit 1K, 2K, and 4K pixel sizes were supported, including
1024x1024,1536x1024,2048x2048,2048x1152,3840x2160, and2160x3840 -
The
gpt-image-2-vipmodel name and $0.03/call per-request billing remain unchanged -
For current 2K / 4K or precise
sizecontrol, use official-forwardgpt-image-2 - 📝 Developer documentation:
Default-group
gpt-image-2 still does not support size / quality. Default-group gpt-image-2-vip is also not a current 2K / 4K size-control route. When you need size and quality parameters, use gpt-image-2 under the Sora2Official / GPTImage2 Enterprise groups.⚠️ sora-2 and sora-2-pro Official-Forward Models Will Be Retired on July 1
June 17, 2026Sora 2 official-forward route retirement notice
-
Due to upstream OpenAI compute allocation changes, the
sora-2andsora-2-proofficial-forward routes have recently experienced severe timeout issues and no longer meet production stability expectations -
LaoZhang API will stop maintaining and retire the
sora-2/sora-2-proofficial-forward models on July 1, 2026 - Although the official retirement timeline may point to September 2026, these routes are already frequently unstable, so we do not recommend new integrations or production workloads
-
Existing users should migrate as early as possible; currently recommended alternatives are
wan-2.7andSeeDance2/ Seedance 2.0 - 📝 Migration reference:
⚠️ gpt-image-2-vip Size Parameter No Longer Supported (Historical Notice)
June 16, 2026gpt-image-2-vip parameter support update
This is a historical notice.
gpt-image-2-vip briefly restored size on June 18, 2026, but the latest state is the June 23, 2026 notice: size is unavailable again and reliable output is currently limited to 1K.- This entry describes the temporary parameter adjustment on June 16, 2026
-
The
gpt-image-2-vipmodel name and $0.03/call per-request billing remain unchanged -
Current status: default-group
gpt-image-2-viplostsizesupport again; submittedsizeis silently ignored, and 2K / 4K cannot be selected -
Default-group
gpt-image-2still does not supportsize/quality -
Use
gpt-image-2under theSora2OfficialorGPTImage2 Enterprisegroup when you need full official Images API parameter compatibility - 📝 Developer documentation:
📅 May 2026
🎬 Veo 3.1 Official API Forwarding Is Live With Low Unified 4K Pricing
May 21, 2026Veo 3.1 official-forward route update
-
🚀 New models:
veo-3.1-fast-generate-previewveo-3.1-generate-preview
-
💵 Pay-per-request billing:
veo-3.1-fast-generate-preview:$0.3/callveo-3.1-generate-preview:$1.2/call- Set Billing mode to
Pay-per-requestwhen creating the token - Supported combinations use one unified price with no resolution surcharge; use
8seconds for production integrations, and1080p/4Krequire8seconds
-
📉 4K pricing advantage:
- Based on Google’s 8-second 4K official price, Fast goes from about
$2.40to$0.3/call - Standard goes from about
$4.80to$1.2/call - 4K requests use unified pricing; verify final resolution from the downloaded MP4 metadata
- Based on Google’s 8-second 4K official price, Fast goes from about
- 📝 Developer documentation:
The new route uses the same OpenAI Videos API style as Sora2 official forwarding:
POST /v1/videos to create a task, GET /v1/videos/{id} to poll status, and GET /v1/videos/{id}/content to download MP4 bytes. The legacy-route incident notice still applies to old model names such as veo-3.1, veo-3.1-fast, and veo-3.1-fl; use the official-forward model names above for new integrations.⚠️ Veo-3.1 Legacy Route Temporarily Unavailable
May 14, 2026Veo-3.1 route incident notice
- The
veo-3.1series legacy route began experiencing failures on May 14, 2026 - The legacy route is temporarily unavailable; pause new calls and production integrations for now
- Existing users should follow later announcements; recovery timing will be announced on this page and in the console
🛡️ GPTImage2 Enterprise Pure Official-Key Group Is Live
May 13, 2026GPT Image 2 official-compatible group update
- New
GPTImage2 Enterprisegroup: pure official-key routing with enterprise concurrency support - Billing is official pricing +20% to cover tax and operating cost, not profit markup
Sora2Officialremains available as a mixed AZ + official-key transit group- Both official-compatible groups use
model="gpt-image-2"with Images Generations and Images Edits
For quick testing or default-group routes, you can keep using default-group
gpt-image-2 / gpt-image-2-vip. For official parameter compatibility and higher stability, choose GPTImage2 Enterprise.🎨 gpt-image-2-vip Size Parameter Historical Notice
May 1, 2026gpt-image-2-vip size capability update
-
✅ Parameter update:
gpt-image-2-vipnow supports 1K, 2K, and 4Ksizetiers30 explicit pixel sizes are supported across square, portrait, landscape, widescreen, and ultrawide ratiosPass an explicit pixel value such as2048x2048,3840x2160, or2160x3840- Current status:
sizeis unavailable again, so 2K / 4K cannot be selected through this route
-
🧩 Integration notes:
gpt-image-2-vipsupportssizegpt-image-2-vipcurrently generates 1K normally; submittedsizeis silently ignored- Use
gpt-image-2under theSora2Officialmixed AZ + official-key group, or use the newGPTImage2 Enterprisepure official-key group when you needquality, 2K / 4K, or full official parameter compatibility
- 📝 Developer documentation:
This historical announcement records the
gpt-image-2-vip size-parameter capability update at that time. It does not change the April 22 announcement for the initial GPT-Image-2 default-group route launch.📅 April 2026
🎨 GPT-Image-2 Default-Group Routes Updated; sora_image Official Route Offline
April 22, 2026Image generation route update
-
🚀 New routes:
gpt-image-2default standard route is now available for text-to-image and image-to-image workflowsgpt-image-2-vipis available; 1K output currently works, whilesizeis temporarily unavailable
-
⚠️ Route change:
- The
sora_imageofficial route is now offline - New image generation integrations should use
gpt-image-2orgpt-image-2-vip - Existing
sora_imagecalls should migrate model and endpoint configuration as soon as possible
- The
-
💵 Pricing:
gpt-image-2is billed per request at$0.03/callgpt-image-2-vipis also billed per request at$0.03/call- You can test the model online at yingtu.ai before integrating the API
- 📝 Developer documentation:
Default-group routes use Images Generations and Images Edits. Use
gpt-image-2 or gpt-image-2-vip for basic text-to-image. For 2K / 4K, precise size, or full official parameter compatibility, use gpt-image-2 under the Sora2Official mixed AZ + official-key group, and choose GPTImage2 Enterprise when you want the pure official-key route.🧠 Claude Opus 4.7 / Thinking Now Available
April 17, 2026Claude Opus 4.7 is now available on LaoZhang API
-
🚀 New Models:
claude-opus-4-7/claude-opus-4-7-thinking -
🌟 Launch Details:
- Standard and Thinking variants are now available
- Works directly through the OpenAI-compatible API by switching the model name
- Suitable for complex reasoning, coding agents, and long-context tasks
-
📝 Documentation Sync:
- Announcement page has been updated
- Related recommendation pages will be updated to the latest model version gradually
To try the latest Claude high-end model, switch the
model field in your existing code or dashboard configuration to either of the model names above.💰 Nano Banana Pricing Update Notice
April 13, 2026New pricing is now live for Nano Banana Pro and Nano Banana2
-
🎯 Affected models:
gemini-3-pro-image-preview(Nano Banana Pro)gemini-3.1-flash-image-preview(Nano Banana2)
-
💵 Price changes:
- Nano Banana Pro:
$0.05/request → $0.09/request - Nano Banana2:
$0.045/request → $0.055/request
- Nano Banana Pro:
-
📝 What was updated:
- Documentation and pricing pages have been synced
- Comparison pages, FAQs, and recommendations now reflect the new pricing
- Actual charges are shown in the console call logs
This update is effective immediately. For bulk purchasing or stable large-scale usage, contact Telegram @laozhang_cn.
📅 January 2026
📚 Documentation Update for 2026
January 7, 2026Documentation Optimization - 2026 New Year Update
- 📅 Year Updates: All model recommendation pages updated to 2026
- 🔄 Model Reference Updates: Default models in code examples updated to latest recommendations
- ⚠️ Retirement Notice: Claude 3 Opus officially retired on January 5, 2026
- 📝 Config Updates: Claude Code configuration method updated to settings.json
New year, new beginnings! LaoZhang API continues to provide you with the best AI services!
📅 December 2025
🚀 GPT-5.2 Stunning Release
December 11, 2025GPT-5.2 - OpenAI’s Annual Flagship Model
-
🚀 New Model:
gpt-5.2/gpt-5.2-2025-12-11 -
💰 Pricing Details:
- Same price as OpenAI official
- Actual charges follow the real-time console price
-
🌟 Core Features:
- 30% less hallucination rate, significantly improved accuracy
- Supports Instant/Thinking/Pro three modes
- 128K ultra-long context window
- Excellent performance in professional work tasks
GPT-5.2 is OpenAI’s 2025 year-end masterpiece, particularly outstanding in professional work scenarios. Recommended for enterprise users requiring high accuracy.
📅 November 2025
🧠 Claude Opus 4.5 Most Intelligent Model Release
November 24, 2025Claude Opus 4.5 - Anthropic’s Strongest Brain
-
🚀 New Model:
claude-opus-4-5-20251101/claude-opus-4-5-20251101-thinking -
💰 Pricing Details:
- Same price as Anthropic official
- Pay-as-you-go usage for better cost control
-
🌟 Core Features:
- Anthropic’s most intelligent model to date
- Supports effort parameter to control response token count
- Top-tier in coding, agents, computer use capabilities
- 200K ultra-long context window
- Supports Thinking chain-of-thought reasoning mode
Claude Opus 4.5 excels in complex reasoning and long document processing, the first choice for enterprise-grade challenging tasks.
🌟 Gemini 3 Series Major Launch
November 18, 2025Gemini 3 Pro/Flash - Google’s Next-Gen AI
-
🚀 New Models:
gemini-3-pro-preview- Flagship version, 2M ultra-long contextgemini-3-flash-preview- Lightning version, new default model for Gemini appgemini-3-pro-image-preview- Image generation specialized
-
💰 Pricing Details:
- Same price as Google official
- Actual charges follow the real-time console price
-
🌟 Core Features:
- Major capability upgrade compared to Gemini 2.5
- Pro version supports 2M token ultra-long context
- Flash version provides lightning-fast response speed
- Supports Thinking/NoThinking dual modes
Gemini 3 series is Google’s latest generation model, now the new default model for Gemini app. Recommended for long text processing and fast response scenarios.
📦 GPT-5.1 Series Launch
November 13, 2025GPT-5.1 Series - Coding Capability Upgraded
-
🚀 New Models:
gpt-5.1/gpt-5.1-2025-11-13- Main versiongpt-5.1-codexseries - Coding specialized (High/Medium/Mini)
-
🌟 Core Features:
- Strong comprehensive capability, balanced cost-performance
- Codex series provides multi-tier coding model selection
- 128K ultra-long context window
GPT-5.1-Codex series is specially optimized for coding scenarios, with corresponding versions from complex architecture design to quick code completion.
📅 October 2025
⚡ Claude Haiku 4.5 Fast Version Launch
October 15, 2025Claude Haiku 4.5 - Speed and Cost Optimized
-
🚀 New Model:
claude-haiku-4-5-20251001/claude-haiku-4-5-20251001-thinking -
💰 Pricing Details:
- Fast response, lower cost
- Suitable for large-scale concurrent calls
-
🌟 Core Features:
- Fast response, extremely low latency
- Supports Thinking reasoning mode
- 200K context window
- First choice for daily quick tasks
Claude Haiku 4.5 is the best choice for cost-sensitive applications, significantly reducing costs while maintaining high-quality output.
📅 September 2025
💻 Claude Sonnet 4.5 Coding Champion Launch
September 29, 2025Claude Sonnet 4.5 - Developer’s First Choice
-
🚀 New Model:
claude-sonnet-4-5-20250929/claude-sonnet-4-5-20250929-thinking -
💰 Pricing Details:
- Balanced performance and cost
- Best value for coding scenarios
-
🌟 Core Features:
- Top coding capability, leading code generation quality
- Suitable as AI agent’s brain
- Outstanding frontend and UI development ability
- Supports Thinking chain-of-thought reasoning
Claude Sonnet 4.5 performs excellently in coding scenarios, particularly suitable as backend model for programming tools like Cursor, Cline.
📅 August 2025
🧠 DeepSeek V3.1 Mixed Reasoning Mode Launch
August 26, 2025DeepSeek V3.1-250821 - Revolutionary Mixed Reasoning Mode
-
🚀 New Model:
deepseek-v3-1-250821 -
💰 Pricing Details:
- Prompt price:
$0.50 / 1M tokens(Official$0.56) - Completion price:
$1.50 / 1M tokens(Official$1.68) - Save about 11% compared to official prices
- Flexible billing for dual mode calls
- Prompt price:
-
🧠 Core Features:
- Think Mode: Deep reasoning with full thought process display
- Non-Think Mode: Quick response for daily tasks
- 128K ultra-long context window
- Anthropic API format compatible
-
🌟 Technical Highlights:
- 840 billion token continuous pre-training optimization
- New tokenizer with improved encoding efficiency
- Beta Function Calling tool invocation support
- Enhanced agent and tool usage capabilities
-
⚡ Mode Switching:
deepseek-chat- Non-Think fast modedeepseek-reasoner- Think reasoning mode- Official site supports “DeepThink” one-click switch
DeepSeek V3.1 is the first large model supporting mixed reasoning mode, allowing users to flexibly choose reasoning depth based on task complexity, balancing efficiency and quality.
🤖 Kimi K2 Official Version Integration
August 25, 2025Kimi K2-250711 - Volcano Engine Official Partnership Version
-
🚀 New Model:
kimi-k2-250711 -
🔄 Replaced Model: Replaces
kimi-k2-0711-previewpreview version -
🤝 Official Partnership:
- Volcano Engine official authorized partnership version
- Direct connection to official interface, significantly improved stability
- Goodbye to preview version instability
-
💰 Pricing Details:
- Continues to maintain 15% lower than official price advantage
- Actual charges follow the real-time console price
- Official version performance at preview version price
-
🌟 Core Features:
- Excellent Chinese understanding and generation capabilities
- Strong long text processing ability
- More stable and reliable API responses
- Fully compatible with original calling methods
Kimi K2-250711 is the official version in partnership with Volcano Engine, with significant improvements in stability and performance compared to preview version. Users are recommended to switch in time.
🚀 GPT-5 Full Series Launch
August 8, 2025GPT-5 Full Series Official Release - OpenAI’s Strongest Model
-
🎯 Full Series Models:
gpt-5- Flagship versiongpt-5-2025-08-07- Specific versiongpt-5-chat-latest- Latest conversation versiongpt-5-mini/gpt-5-mini-2025-08-07- Lightweight versiongpt-5-nano/gpt-5-nano-2025-08-07- Ultra-lightweight version
-
💰 Pricing Details:
- Exactly same price as OpenAI official site
- Actual charges follow the real-time console price
- Prompt price:
$1.25 - $0.05 / 1M tokens - Completion price:
$10.00 - $0.40 / 1M tokens
-
🛠️ Technical Features:
- Supports official documentation
/v1/responsesendpoint calls - Temperature parameter must be set to 1 or not passed (official restriction)
- Use
max_completion_tokensinstead ofmax_tokens gpt-5-chat-latestcalled through/v1/chat/completions
- Supports official documentation
-
⚡ Performance Highlights:
- Comprehensive reasoning capabilities surpassing GPT-4 series
- Stronger context understanding and long text processing
- Significantly improved code generation quality
- Multilingual capabilities reaching new heights
🎨 Claude Opus 4.1 Performance Upgrade Launch
August 7, 2025Claude Opus 4.1 - Code Capability Breakthrough
-
🚀 New Model:
claude-opus-4-1-20250805 -
💰 Increased Capacity Without Price Increase:
- Price exactly same as Anthropic official site
- Pay-as-you-go usage for better cost control
- Same price, stronger performance
-
🌟 Core Upgrades:
- SWE-bench Verified reaches 74.5%, significantly improved coding ability
- Compared to Opus 4, reasoning ability and code generation quality greatly enhanced
- Maintains 200K ultra-long context window
- Further optimized multilingual understanding and generation
-
⚡ Performance Highlights:
- Code debugging and fixing capabilities leading the industry
- Complex reasoning task accuracy improved by 15%+
- API response speed maintains industry-leading level
- Perfectly compatible with all Claude series features
Claude Opus 4.1 is an iterative optimization based on Opus 4, maintaining the original powerful capabilities while making special optimizations for code generation and reasoning tasks.
🎯 OpenAI Open Source Models Official Launch
August 6, 2025OpenAI’s First Open Source Large Model Release
-
🚀 New Models:
gpt-oss-120b- 117 billion parameters, comparable to o4-mini performancegpt-oss-20b- 21 billion parameters, can run on edge devices
-
💰 Pricing Details:
- Priced significantly lower than DeepSeek R1 and V3
- Enterprise usage costs significantly reduced
-
🌟 Core Features:
- 128K ultra-long context window
- Supports low/medium/high three-level reasoning modes
- MoE architecture, extremely efficient inference
- Apache 2.0 open source license
-
⚡ Performance Highlights:
- gpt-oss-120b only needs a single 80GB GPU to run
- gpt-oss-20b supports deployment on 16GB memory devices
- Excellent performance in coding, math, tool calling tasks
This is OpenAI’s first open source model release since GPT-2, marking an important milestone in the AI open ecosystem.
📅 July 2025
⚠️ GPT-4.5-Preview Model Discontinued
July 14, 2025GPT-4.5-Preview Officially Discontinued
- 📌 Discontinued model:
gpt-4.5-preview - 🔄 Recommended alternative:
gpt-4.1 - 📅 Effective date: July 14, 2025
- 📖 Official announcement: OpenAI Deprecations
🚀 Kimi K2 Model Launch
July 15, 2025Kimi K2 Preview Version Launch
- 📌 Model name:
kimi-k2-0711-preview - 💰 Price advantage: 15% lower than official price
- 📊 Actual charges follow the real-time console price
- 🌟 Features: Better Chinese understanding and generation capabilities
🤖 Grok-4 Model Launch
July 12, 2025xAI Grok-4 Official Integration
- 📌 Call name:
grok-4 - 🧠 Special feature: Built-in reasoning mode
- 📖 Recommended to check official documentation for detailed usage
💸 Claude 20% Price Reduction
July 8, 2025Claude Models Significant Discount
- 🎉 All series models reduced by 20%
- ✅ No speed limit, no account ban
- 📊 Actual charges follow the real-time console price
- 🚀 Providing more economical AI services to users
📅 June 2025
🌟 Gemini 2.5 Official Release
June 18, 2025Gemini 2.5 Series All-new Upgrade
- 📌 Recommended models:
gemini-2.5-pro/gemini-2.5-flash - ⚡ More stable performance, recommended to replace preview or exp versions
- 🎯 2M ultra-long context support
🎊 O3 Model Massive 80% Price Reduction
June 13, 2025O3 Model Largest Price Drop in History
- 💥 Price reduction: 80%
- 💰 New price:
- Input:
$3/M Tokens - Output:
$12/M Tokens
- Input:
- 🆕 New
o3-promodel added - 📍 Only callable via
/v1/responsesendpoint
🖼️ GPT-Image-1 Price Reduction
June 7, 2025More Affordable Image Generation
- 📌 Model:
gpt-image-1 - 💸 Significant price reduction, providing more competitive image generation services
- 🎨 Reduced costs while maintaining high-quality output
🔔 Subscribe to Updates
Join Community
Join Telegram community for real-time update notifications
View Pricing
Check latest pricing for all models
📊 Update Statistics
This Month’s Highlights (June 2026)
- ⚠️ Video route update:
sora-2andsora-2-proofficial-forward models will stop being maintained and be retired on July 1, 2026 - 🧭 Migration guidance: users on Sora 2 official-forward routes should move early to
wan-2.7orSeeDance2/ Seedance 2.0 - 🎨 Parameter update:
gpt-image-2-viplostsizeagain on June 23; 1K output works, while 2K / 4K cannot be selected through this route for now - 🧩 Integration reminder: default-group
gpt-image-2still does not supportsize/quality; do not use default-groupgpt-image-2-vipfor 2K / 4K size control right now - 🛡️ Migration path: use
gpt-image-2under theSora2OfficialorGPTImage2 Enterprisegroup when you need precisesize, 2K / 4K, or full official parameter compatibility - 📝 Announcement sync: the Sora 2 official-forward retirement notice and GPT Image 2 parameter update are now reflected in both Chinese and English
Historical Milestones
- June 2026:
sora-2andsora-2-proofficial-forward models were announced for July 1 retirement;gpt-image-2-vipbriefly restoredsize, then lost it again on June 23, with reliable output currently limited to 1K - May 2026: Veo 3.1 official API forwarding launched with low unified 4K pricing; the
veo-3.1series legacy route became temporarily unavailable from May 14;gpt-image-2-vipadded 1K/2K/4K size capability;GPTImage2 Enterprisewas added as a pure official-key group - April 2026: GPT-Image-2 default-group routes launched;
sora_imageofficial route went offline; Claude Opus 4.7 / Thinking launched; Nano Banana series pricing updated - January 2026: Documentation fully updated for new year; Claude 3 Opus officially retired
- December 2025: GPT-5.2 stunning release, 30% less hallucination
- November 2025: Claude Opus 4.5 most intelligent model released; Gemini 3 series launched; GPT-5.1 series launched
- October 2025: Claude Haiku 4.5 fast version launched
- September 2025: Claude Sonnet 4.5 coding champion launched
- August 2025: DeepSeek V3.1 mixed reasoning mode; Kimi K2 official version; GPT-5 full series; OpenAI open source models
- July 2025: Surpassed 200+ model support
- June 2025: O3 model largest price reduction in history
- May 2025: Gemini 2.5 series support
Update Frequency: This page is updated regularly. Recommended to bookmark and check frequently. All price adjustments and model updates will be announced here first.