Skip to main content
This archive lists the LaoZhang API announcements published in 2026, newest first. Each entry records the facts at the time of publication; when a later change superseded an entry, the entry says so. Current models and prices are in the model and pricing catalog.

​
October 2026

​
October 14, 2026: New Seedance prices

  • The SeeDance2 group multiplier moves from 0.18 to 0.1714, so Seedance 2.0 and 2.5 cost 1.2× Volcengine’s list price, and most rates drop about 5%.
  • Seedance 2.5 at 1080p now uses Volcengine’s 1080p rate: $13.20 per 1M tokens without an input video and $7.88 with one.
  • Read the announcement

​
October 14, 2026: Wan 3.0 launch, new Wan video prices, and Wan 2.6 retirement

  • wan3.0-video (standard) and wan3.0-video-prime (faster) are available in the Wan group, billed per second of video at 480P, 720P, or 1080P.
  • Wan video now costs Alibaba Cloud’s list price × 0.1714 in USD per second (about 1.2×). Wan 2.7 moves from $0.09 at 720P and $0.15 at 1080P to $0.1028 and $0.1714.
  • wan2.6-t2v, wan2.6-i2v, wan2.6-r2v, and wan2.6-r2v-flash are retired.
  • Read the announcement

​
October 14, 2026: Gemini 4 Argon launch

  • gemini-4-argon, announced by Google on September 30, is available in the default group with usage-based billing through the OpenAI-compatible and Gemini-native endpoints. It costs $2 input and $10 output per 1M tokens, the same as Google’s launch price, and thinking tokens are billed as output.
  • Thinking levels are low, medium, and high. Google says the price rises to $4 input and $20 output after the launch period, without giving an end date.
  • Read the launch announcement

​
October 13, 2026: GPT-6.1 Sol launch

  • gpt-6.1-sol, released by OpenAI on September 29, is available in the default group with usage-based billing through Chat Completions and Responses. Standard requests cost $2 input, $10 output, and $0.1 cache read per 1M tokens, the same as OpenAI.
  • Chat Completions doesn’t support function calling, and reasoning_effort accepts low through xhigh; tool calling and max require the Responses API.
  • Read the launch announcement

​
October 7, 2026: Nano Banana 2.1 launch

  • LaoZhang API added Google’s Nano Banana 2.1 (gemini-nano-banana-2.1) in the default group at $0.045 per call, the same for 1K, 2K, and 4K. It supports 14 aspect ratios, up to 14 references, and web and image search grounding.
  • It takes the same requests as Nano Banana 2 ($0.055 per call), so switching means changing the model ID. Nano Banana 2.1 doesn’t accept 512 (0.5K), and its default thinking level is medium.
  • Nano Banana 2 stays available at the same price.
  • See what changed and how to switch

​
October 5, 2026: Nano Banana Standard still available for now

  • Google lists October 2, 2026 as the shutdown date for gemini-2.5-flash-image. As of October 5, LaoZhang API still serves the model at $0.02 per call. This supersedes the September 24 note that LaoZhang API would stop it on the same day.
  • There is no guaranteed end date, and the model can stop at any time. Once it stops, requests won’t move to a newer model on their own.
  • See how to migrate

​
September 2026

​
September 25, 2026: Seedream 5.0 Flash and 5.0 Pro launched

  • seedream-5-0-flash-260915 ($0.018 per call) and seedream-5-0-pro-260628 ($0.12 per call) became available on September 25, 2026 in the default group, on the same POST /v1/images/generations endpoint as other Seedream models. Each output image is charged once. Flash doesn’t charge for reference images; Pro charges $0.003 for each reference image after the first.
  • Neither model supports sequential generation or streaming. 5.0 Flash outputs up to 2K and can split an image into transparent layers, and 5.0 Pro outputs 1K or 2K.
  • Read the launch announcement

​
September 24, 2026: GPT-6 Astra, Sol, and Luna launched

  • gpt-6-astra went live on LaoZhang API on September 10, 2026, and gpt-6-sol and gpt-6-luna were available in the pricing configuration as of September 24, 2026. All three use the default group and the OpenAI-compatible Chat Completions or Responses endpoints, with usage-based billing.
  • For requests with up to 272K input tokens, input and output cost $10 and $50 per 1M tokens for gpt-6-astra, $2 and $10 for gpt-6-sol, and $0.1 and $0.5 for gpt-6-luna, the same as OpenAI’s prices. A request with more than 272K input tokens is billed entirely at long-context rates.
  • Read the launch announcement

​
September 24, 2026: GPT-5.6 Sol prices reduced

  • OpenAI cut gpt-5.6-sol prices on August 21, 2026, and LaoZhang API updated its pricing configuration to the new rates on September 10. Standard requests went from $5 to $4 per 1M input tokens and from $30 to $20 per 1M output tokens; cache reads, cache writes, and long-context rates above 272K input tokens fell as well.
  • OpenAI describes the new rates as promotional pricing available at least through November 21, 2026. Existing API keys, base URLs, model IDs, and request parameters did not change.
  • Read the price cut notice

​
September 24, 2026: Sora 2 video API retired

​
September 24, 2026: API domains

  • The default API domain is api.laozhang.ai. Servers in Europe or North America can connect directly through api-vip.laozhang.ai, and api2.laozhang.ai is the fallback when api.laozhang.ai can’t be reached.
  • All three accept the same API key, paths, and request parameters, so switching changes only the hostname.
  • See the domain setup

​
September 24, 2026: Nano Banana Standard shutdown

  • Google shuts down gemini-2.5-flash-image on October 2, 2026, and LaoZhang API stops serving it at the same time. Move existing integrations to Nano Banana 2 (gemini-3.1-flash-image).
  • gemini-3.1-flash-image-preview and gemini-3-pro-image-preview are mapped to the matching stable IDs.
  • See how to migrate
  • Superseded by the October 5 entry: LaoZhang API kept serving the model after October 2.

​
September 24, 2026: GPT-Image-1 shutdown

  • OpenAI removes gpt-image-1 from its API on October 23, 2026, and gpt-image-1.5 and gpt-image-1-mini on December 1, 2026. LaoZhang API stops serving each model at the same time.
  • Use GPT Image 2.5 for new integrations.
  • See the GPT Image 2.5 guide

​
September 22, 2026: GPT Image 2.5 VIP models restored

  • gpt-image-2.5-flare-vip and gpt-image-2.5-sunburst-vip became available again. Requests resumed with a per-call token in the default group.
  • Read the recovery notice

​
September 16, 2026: GPT Image 2.5 VIP models paused

  • Both GPT Image 2.5 -vip models were paused because of limited upstream capacity. The official-forward models and gpt-image-2.5-web stayed available. Both -vip models were restored on September 22.
  • From September 16, the web 2.5 route uses the model ID gpt-image-2.5-web.
  • Read the incident record

​
September 9, 2026: GPT Image 2.5 launched

  • GPT Image 2.5 Flare (gpt-image-2.5-flare) and Sunburst (gpt-image-2.5-sunburst) launched through official API forwarding, with token billing at the official-forward gpt-image-2 rates, in these groups:
    • Sora2Official
    • GPTImage2 Sora2 Enterprise
  • gpt-image-2.5-flare-vip and gpt-image-2.5-sunburst-vip launched in the default group at $0.03 per call each.
  • At launch, the announcement directed default group web-route users to gpt-image-2-web.
  • Correction: since September 16, 2026, web version 2.5 uses gpt-image-2.5-web. gpt-image-2-web is a legacy web ID on a separate route, and parameters for one do not carry over to the other.
  • Read the launch announcement

​
September 3, 2026: Gemini 3.8 Flash launched

  • gemini-3.8-flash launched with usage-based billing through the OpenAI-compatible Chat Completions API. The console listed $0.75 per 1M input tokens and $3.75 per 1M output tokens at launch.
  • Read the launch announcement

​
September 3, 2026: Grok Imagine Image 2.0 launched

  • grok-imagine-image-2.0 launched at $0.055 per successful output image, with 1K/2K output, 15 fixed aspect ratios, up to 10 outputs per request, and multipart editing with one to three references.
  • Read the launch announcement

​
September 3, 2026: Legacy Grok Imagine price change

​
August 2026

​
August 13, 2026: Grok Imagine image models launched

  • grok-imagine-image launched at $0.02 per image and grok-imagine-image-quality at $0.045 per image, with five documented aspect ratios, 1K/2K output, and up to 10 outputs per request. grok-imagine-image accepted one to four reference images and grok-imagine-image-quality one or two.
  • Both became legacy models when grok-imagine-image-2.0 launched on September 3.
  • Read the legacy Grok Imagine model notice

​
July 2026

​
July 30, 2026: GPT-5.6 Luna and Terra prices reduced

  • gpt-5.6-luna input, output, and cached-input prices fell 80%, and gpt-5.6-terra prices fell 20%. The new rates applied automatically, with no change to API keys, base URLs, or model IDs.
  • At publication, gpt-5.6-sol pricing was unchanged. Sol prices were cut later; see the September 24 entry.
  • Read the pricing announcement

​
July 23, 2026: Default API domain switched to api2.laozhang.ai

  • The default API domain switched to api2.laozhang.ai, with api-cf.laozhang.ai and api-vip.laozhang.ai as fallback routes. API keys, paths, and request parameters did not change.
  • Updated on September 24, 2026: the default domain is api.laozhang.ai again, and api2.laozhang.ai is the fallback when it can’t be reached; see the September 24 “API domains” entry above.
  • Read the migration guide

​
July 22, 2026: Gemini 3.6 Flash and Gemini 3.5 Flash-Lite launched

  • gemini-3.6-flash and gemini-3.5-flash-lite launched through the OpenAI-compatible API. Gemini 3.8 Flash followed on September 3.
  • Read the launch announcement

​
July 22, 2026: gpt-image-2-vip 4K and Nano Banana route status

  • gpt-image-2-vip supported 4K sizes and the quality parameter again, and the Nano Banana Pro, Nano Banana 2, and Nano Banana 2 Lite routes were reported stable for high-concurrency production use.
  • Read the status announcement

​
July 9, 2026: gpt-image-2-vip size and quality restored

  • gpt-image-2-vip in the default group supported these parameters again. This superseded the June 23 notice.
    • size: common 1K, 2K, and 4K outputs
    • quality: low, medium, and high
  • For stricter official parameter compatibility, gpt-image-2 remained available in Sora2Official and GPTImage2 Sora2 Enterprise with usage-based billing.
  • GPT Image 2 API guide

​
July 1, 2026: Nano Banana 2 Lite launched

  • gemini-3.1-flash-lite-image (Nano Banana 2 Lite) launched with per-call billing only, at $0.025 per call, for 1K output in 14 aspect ratios.
  • Nano Banana 2 Lite API guide

​
July 1, 2026: Claude Sonnet 5 launched

  • claude-sonnet-5 launched with introductory pricing that followed Anthropic’s: $2 input and $10 output per 1M tokens through August 31, 2026, then $3 and $15.
  • Claude models have not been listed on LaoZhang API since September 24, 2026. We will post an announcement if they return.
  • Claude API guide

​
June 2026

​
June 23, 2026: gpt-image-2-vip size unavailable again

  • After an upstream rule change, gpt-image-2-vip in the default group ignored size again and returned 1K images. 2K, 4K, and exact sizes required official-forward gpt-image-2.
  • Superseded by the July 9 restoration.

​
June 18, 2026: gpt-image-2-vip size briefly restored

  • gpt-image-2-vip in the default group briefly supported explicit 1K, 2K, and 4K sizes again, such as 2048x2048 and 3840x2160.
  • Superseded by the June 23 notice.

​
June 17, 2026: Sora 2 official-forward retirement notice

  • Because of severe upstream timeouts, LaoZhang API announced that it would stop maintaining and retire the sora-2 and sora-2-pro official-forward models on July 1, 2026. It recommended Wan 2.7 models such as wan2.7-t2v and wan2.7-i2v, or Seedance 2.0.
  • Outcome: OpenAI removed its Videos API and both Sora 2 models from its API on September 24, 2026. See the September 24 entry.
  • Wan 2.7 video API · Seedance 2.0 video API

​
June 16, 2026: gpt-image-2-vip size support removed

  • gpt-image-2-vip in the default group stopped honoring size, so 2K and 4K could not be selected.
  • Full official Images API parameters required gpt-image-2 in Sora2Official or GPTImage2 Sora2 Enterprise.
  • Superseded by the June 18 notice.

​
May 2026

​
May 21, 2026: Veo 3.1 official API forwarding launched

  • veo-3.1-fast-generate-preview launched at $0.3 per call and veo-3.1-generate-preview at $1.2 per call, using the /v1/videos task flow. 1080p and 4K required 8-second videos.
  • No Veo models are in the current public model catalog. Check availability in the console.

​
May 14, 2026: Veo 3.1 legacy route unavailable

  • The legacy Veo 3.1 route (veo-3.1, veo-3.1-fast, veo-3.1-fl) began failing, and new calls and production use were paused.
  • No Veo models are in the current public model catalog. Check availability in the console; for video generation, see Wan 2.7 and Seedance 2.0.

​
May 13, 2026: Enterprise official-key group for GPT Image 2 launched

  • A pure official-key group for gpt-image-2 launched with enterprise concurrency support. The console now shows it as GPTImage2 Sora2 Enterprise.
  • It was billed at the official price plus 20% to cover upstream and operating costs. Sora2Official remained available as a mixed AZ and official-key group.
  • GPT Image 2 API guide

​
May 1, 2026: gpt-image-2-vip size parameter added

  • gpt-image-2-vip in the default group added 1K, 2K, and 4K size support with 30 explicit pixel sizes.
  • Example sizes: 2048x2048, 3840x2160, and 2160x3840.
  • Later changes: size stopped working on June 16, returned briefly on June 18, stopped again on June 23, and returned with quality on July 9.

​
April 2026

​
April 22, 2026: GPT Image 2 launched in the default group

  • gpt-image-2 and gpt-image-2-vip launched in the default group for text-to-image and image-to-image, each at $0.03 per call.
  • The sora_image official route went offline, and existing sora_image integrations were asked to migrate to gpt-image-2 or gpt-image-2-vip.
  • GPT Image 2 API guide

​
April 17, 2026: Claude Opus 4.7 launched

  • claude-opus-4-7 and claude-opus-4-7-thinking launched through the OpenAI-compatible API.
  • Claude models have not been listed on LaoZhang API since September 24, 2026. We will post an announcement if they return.

​
April 13, 2026: Nano Banana Pro and Nano Banana 2 price change

  • gemini-3-pro-image-preview (Nano Banana Pro) changed from $0.05 to $0.09 per call, and gemini-3.1-flash-image-preview (Nano Banana 2) from $0.045 to $0.055 per call.
  • Nano Banana Pro API · Nano Banana 2 API

​
January 2026

​
January 7, 2026: Documentation updated for 2026

  • Model recommendation pages and code-example defaults were updated, and the Claude Code configuration guide moved to settings.json. The update also noted that Claude 3 Opus was retired on January 5, 2026.
Back to announcements