Developer Dashboard

Deprecated Models

This page contains information about models that have been deprecated by their respective providers and are no longer available through AvalAI.

Overview

When AI model providers deprecate models, they are no longer available for new requests. This page serves as a comprehensive reference for deprecated models and provides guidance on migration to supported alternatives.

Deprecation Lifecycle

OpenAI's official deprecation guidance distinguishes between legacy, deprecated, and shut down models or endpoints:

  • Legacy means the model or endpoint no longer receives active updates and should be treated as a migration candidate.
  • Deprecated means the provider has announced retirement and published a shutdown or sunset date.
  • Shut down means requests will no longer be served after the listed date.

OpenAI generally gives at least 6 months of notice for generally available models, at least 3 months for specialized GA variants such as chat, Codex, or deep-research snapshots, and much shorter notice for preview models. Other AvalAI providers may use different notice windows, so always treat provider announcements, this page, and the live /v1/models catalog as the source of truth.

Warning

If OpenAI recommends a replacement that is not yet enabled in AvalAI, choose the nearest supported equivalent from Model Selection, the provider page, or /v1/models instead of hard-coding an unavailable model ID.

Migration Playbook

Use this checklist before a shutdown date:

  1. Inventory usage: search code, prompts, config, queues, eval fixtures, and saved presets for deprecated IDs and aliases.
  2. Select supported replacements: prefer stable, non-preview model IDs in AvalAI; avoid -latest aliases for regulated or regression-sensitive workloads.
  3. Run side-by-side evals: compare quality, latency, cost, context-window usage, tool behavior, structured-output validity, and safety behavior.
  4. Ship behind controls: roll out through feature flags or per-customer configuration so you can revert quickly.
  5. Monitor the cutover: alert on 404, model_not_found, provider-routing errors, cost changes, and unexpected usage deltas.

Important Note on Model Naming

Best Practice: Always use stable model namespaces when they become available. When a preview model is first introduced (e.g., gemini-2.5-flash-image-preview), it may later be released as a stable version (e.g., gemini-2.5-flash-image). In such cases, it is strongly advised to migrate to the non-preview, stable version as soon as possible to ensure continued support and optimal performance.

Upcoming Deprecations

Moonshot.ai Legacy Models

Moonshot.ai is deprecating the following legacy Kimi and Moonshot model IDs:

  • kimi-k2-thinking
  • kimi-k2-0711-preview
  • kimi-thinking-preview
  • kimi-latest-8k
  • kimi-latest-32k
  • kimi-latest-128k
  • moonshot-v1-auto
  • moonshot-v1-8k-vision-preview
  • moonshot-v1-8k
  • moonshot-v1-32k-vision-preview
  • moonshot-v1-32k
  • moonshot-v1-128k-vision-preview
  • moonshot-v1-128k

Migration Path: Migrate to a currently supported Moonshot.ai model before these IDs are removed. Choose an appropriate replacement from the Moonshot.ai models page and verify availability through the live /v1/models catalog.

ElevenLabs Legacy Models (Deprecation Date: July 9, 2026)

ElevenLabs has announced that the following legacy models will be deprecated and removed from the API on July 9, 2026:

Shutdown dateDeprecated modelRecommended replacement
Jul 9, 2026scribe_v1scribe_v2
Jul 9, 2026eleven_monolingual_v1eleven_turbo_v2 or eleven_flash_v2 (English TTS)
Jul 9, 2026eleven_multilingual_v1eleven_multilingual_v2, eleven_turbo_v2_5, or eleven_flash_v2_5 (multilingual TTS)

Models Affected:

  • scribe_v1 - Legacy transcription model (basic speech-to-text)
  • eleven_monolingual_v1 - Legacy English text-to-speech model
  • eleven_multilingual_v1 - Legacy multilingual text-to-speech model

Migration Path: Please migrate to the latest supported ElevenLabs models before July 9, 2026:

  • scribe_v2 - State-of-the-art speech recognition with diarization, word-level timestamps, and entity detection (recommended replacement for scribe_v1)
  • eleven_turbo_v2 / eleven_flash_v2 - High-quality, low-latency English TTS (recommended replacements for eleven_monolingual_v1)
  • eleven_multilingual_v2 / eleven_turbo_v2_5 / eleven_flash_v2_5 - Modern multilingual TTS models (recommended replacements for eleven_multilingual_v1)

For details on the replacement models, see the ElevenLabs Models page.

Groq PlayAI TTS Models (Deprecated: December 31, 2025)

Groq has deprecated the PlayAI TTS models. These models are no longer available:

  • groq.playai-tts - PlayAI Text-to-Speech model
  • groq.playai-tts-arabic - PlayAI Arabic Text-to-Speech model

Important: Requests using these models will fail. Please migrate immediately.

Migration Path: Please migrate to alternative TTS models available through AvalAI.

Gemini 2.5 TTS Preview Models (Update Date: December 10, 2025)

Google has announced that the Gemini Text-to-Speech (TTS) preview models will be updated in place effective December 10, 2025.

Models Affected:

  • gemini-2.5-flash-preview-tts
  • gemini-2.5-pro-preview-tts

What's Changing: Starting on this date, these models will automatically receive significant improvements in expressivity, pacing, and overall audio quality. The new models maintain many characteristics of the previous version to ensure a seamless transition.

Action Required: No action is required. No code changes needed - your existing API calls will automatically begin using the updated model.

Alibaba Qwen Legacy Snapshot Models (Deprecation Date: May 13–31, 2026)

The Bailian Large Model Service Platform (Alibaba DashScope) will be decommissioning several historical snapshot model families. The decommissioning process begins at 00:00:00 (UTC+8) on May 13, 2026 and will be fully completed by May 31, 2026.

During this window, API calls to these models may experience timeouts, failures, or no response. We strongly recommend completing migration before May 13, 2026 to avoid service disruption.

Models Affected (all variants: base alias, -latest, and dated snapshots):

  • qwen-max series - qwen-max, qwen-max-latest, qwen-max-2025-01-25
  • qwen-turbo series - qwen-turbo, qwen-turbo-latest, qwen-turbo-2025-04-28, qwen-turbo-2024-11-01
  • qwen-vl-max series - qwen-vl-max, qwen-vl-max-latest, qwen-vl-max-2025-08-13, qwen-vl-max-2025-04-08
  • qwen-vl-plus series - qwen-vl-plus, qwen-vl-plus-latest, qwen-vl-plus-2025-08-15, qwen-vl-plus-2025-07-10, qwen-vl-plus-2025-05-07, qwen-vl-plus-2025-01-25
  • qvq-max series - qvq-max, qvq-max-latest, qvq-max-2025-03-25
  • Open-source series - qwen2.5-vl-32b-instruct, qwen2.5-vl-72b-instruct, qwen2.5-vl-7b-instruct, qwen2.5-vl-3b-instruct, qwen2.5-7b-instruct-1m, qwen2.5-14b-instruct-1m, qwen2.5-72b-instruct, qwen2.5-32b-instruct, qwen2.5-14b-instruct, qwen2.5-7b-instruct, qwen3-0.6b, qwen3-1.7b, qwen3-4b

Migration Path: Please migrate to the latest supported Qwen models:

  • qwen3.7-max - Latest flagship agent foundation for advanced coding and reasoning (recommended replacement for qwen-max series)
  • qwen3.7-plus - Balanced Qwen3.7 model for high-volume coding, multimodal reasoning, and agentic workloads
  • qwen3.6-plus - Still-supported 1M-context model with agentic coding and multimodal capabilities
  • qwen3.6-flash - Fast, cost-effective replacement for qwen-turbo series
  • qwen3.6-max-preview - 1T+ parameter premium preview model for demanding tasks
  • qwen3-vl-plus / qwen3-vl-flash - Modern vision-language replacements for qwen-vl-max, qwen-vl-plus, and qvq-max series

For details on the replacement models, see the Alibaba Models page.

DeepSeek Chat & Reasoner Models (Deprecation Date: July 24, 2026)

DeepSeek has announced that the legacy model aliases deepseek-chat and deepseek-reasoner will be fully retired on July 24, 2026, 15:59 (UTC). For compatibility, these model names are now routed to the new DeepSeek V4 flagship family: deepseek-chatdeepseek-v4-flash (non-thinking mode) and deepseek-reasonerdeepseek-v4-pro (thinking mode).

Models Affected:

  • deepseek-chat - Legacy alias, now routes to deepseek-v4-flash (pricing unchanged)
  • deepseek-reasoner - Legacy alias, now routes to deepseek-v4-pro (pricing raised to match deepseek-v4-pro)

Migration Path: Please migrate to the explicit DeepSeek V4 models before July 24, 2026:

  • deepseek-v4-flash - Recommended replacement for deepseek-chat (fast, economical flagship with 1M context, supports both thinking and non-thinking modes)
  • deepseek-v4-pro - Recommended replacement for deepseek-reasoner (most capable flagship with open-source SOTA Agentic Coding, 1M context, thinking mode by default)

Gemini 2.5 Flash Image Preview (Deprecation Date: January 15, 2026)

Google has announced that Gemini 2.5 Flash Image Preview (gemini-2.5-flash-image-preview) in the Gemini API will be discontinued on January 15, 2026.

Migration Path: Please migrate to the following generally available model for improved performance and capabilities:

  • gemini-2.5-flash-image - Stable version of Gemini 2.5 Flash image generation

Gemini 3 Pro Preview (gemini-3-pro-preview) (Deprecation Date: March 9, 2026)

Google has deprecated Gemini 3 Pro Preview (gemini-3-pro-preview) on Gemini API and Google AI Studio (AIS), effective March 9, 2026.

Deprecated Model:

  • gemini-3-pro-preview - Gemini 3 Pro Preview model

Key Changes and Timelines:

  • March 6, 2026: If you are using the -latest alias, it switched to Gemini 3.1 Pro Preview (gemini-3.1-pro-preview).
  • March 9, 2026: Gemini 3 Pro Preview was discontinued in favor of Gemini 3.1 Pro Preview.

Migration Path: Please migrate to the following model to avoid service disruption:

  • gemini-3.1-pro-preview - Latest Gemini Pro Preview model

Gemini 2.5 Flash Lite Preview 09-2025 (Deprecation Date: March 31, 2026)

Google has announced that Gemini 2.5 Flash Lite Preview 09-2025 will be discontinued on Gemini API and Google AI Studio (AIS) effective March 31, 2026. Please note that this deprecation only applies to AI Studio and the Gemini API; the model is not being discontinued on Vertex AI.

Key Changes starting March 31, 2026:

  • Gemini 2.5 Flash Lite Preview 09-2025 will be discontinued in favor of Gemini 3.1 Flash Lite Preview.
  • The -latest alias will automatically point to Gemini 3.1 Flash Lite Preview (gemini-3.1-flash-lite-preview).

Migration Path: Please migrate to the following model:

  • gemini-3.1-flash-lite-preview - Latest Gemini Flash Lite Preview model

Anthropic Claude 4 Models (Deprecation Date: June 15, 2026)

Anthropic has announced that the following Claude 4 models will be deprecated on June 15, 2026:

Models Affected:

  • anthropic.claude-sonnet-4-20250514-v1:0 - Claude Sonnet 4 (May 14, 2025 version)
  • anthropic.claude-opus-4-20250514-v1:0 - Claude Opus 4 (May 14, 2025 version)

Migration Path: Please migrate to the latest Claude models before June 15, 2026:

  • claude-sonnet-4-6 - Latest Claude Sonnet model
  • claude-opus-4-8 - Latest Claude Opus model

Gemini 2.0 Flash & Flash Lite (Deprecation Date: June 1, 2026)

Google has announced that the following GA models will be discontinued on Gemini API and Google AI Studio (AIS) on June 1, 2026:

ModelEndpoint IDRequired Action
Gemini 2.0 Flashgemini-2.0-flash, gemini-2.0-flash-001Migrate to a supported Gemini model (e.g., 2.5 Flash, 2.5 Flash Lite)
Gemini 2.0 Flash Litegemini-2.0-flash-lite, gemini-2.0-flash-lite-001Migrate to a supported Gemini model (e.g., 2.5 Flash Lite)

Migration Path: Please migrate to the following models:

  • gemini-2.5-flash - Recommended replacement for Gemini 2.0 Flash
  • gemini-2.5-flash-lite - Recommended replacement for Gemini 2.0 Flash Lite

Imagen 4 Models (Discontinuation Date: August 17, 2026)

Google has announced that the following Imagen 4 model endpoints will be discontinued on August 17, 2026:

  • imagen-4.0-generate-001 - Imagen 4 Standard
  • imagen-4.0-ultra-generate-001 - Imagen 4 Ultra
  • imagen-4.0-fast-generate-001 - Imagen 4 Fast

Important: After the discontinuation date, any API calls sent to these legacy endpoints will fail and return a 404 - Not Found error. The newer Gemini image generation models provide higher capabilities and better performance at the same cost.

Recommended Migration Paths:

Endpoint to be discontinuedRecommended migration path
imagen-4.0-generate-001gemini-3.1-flash-image
imagen-4.0-ultra-generate-001gemini-3.1-flash-image
imagen-4.0-fast-generate-001gemini-3.1-flash-image

Action Required: To avoid service disruptions, you must update your code to reference the new model IDs before August 17, 2026:

  1. Review your current implementation to identify any calls to the legacy endpoints listed above.
  2. Update your API requests to point to the new Gemini 3.1 Flash Image model (gemini-3.1-flash-image).
  3. Test your integration with the new model to ensure a seamless transition for your users.

OpenAI Deprecations

The following schedules reproduce OpenAI's published upstream deprecation data. Availability through AvalAI can differ from direct OpenAI availability, so verify a replacement in the live /v1/models catalog before migration.

2026-06-11: GPT-5 and o3 model deprecations

OpenAI notified developers using older GPT-5 and o3 snapshots that the following models will be removed from its API on December 11, 2026.

Shutdown dateModel / systemRecommended replacement
Dec 11, 2026gpt-5-2025-08-07gpt-5.5
Dec 11, 2026gpt-5-mini-2025-08-07gpt-5.4-mini
Dec 11, 2026gpt-5-nano-2025-08-07gpt-5.4-nano
Dec 11, 2026gpt-5-pro-2025-10-06gpt-5.5-pro
Dec 11, 2026o3-2025-04-16gpt-5.5
Dec 11, 2026o3-pro-2025-06-10gpt-5.5-pro

2026-06-03: Reusable prompts

Reusable prompt objects in the OpenAI dashboard and API are being deprecated.

DateUpdate
June 3, 2026Deprecation announced and prompt creation de-emphasized in the platform.
Nov 30, 2026The v1/prompts API and reusable prompt objects are scheduled to shut down.

Move reusable prompt content into application code. See Migrate from prompt objects.

2026-06-03: Evals platform

OpenAI's Evals platform, including graders documented for eval workflows, is being deprecated. Fine-tuning timelines are listed separately below.

DateUpdate
June 3, 2026Deprecation announced for the Evals platform.
Oct 31, 2026Existing evals become read-only.
Nov 30, 2026The Evals dashboard and API are scheduled to shut down.

See Moving from OpenAI Evals to Promptfoo for a migration path.

2026-06-03: Agent Builder

Agent Builder is being deprecated; ChatKit remains available.

DateUpdate
June 3, 2026Deprecation announced for Agent Builder.
Nov 30, 2026Agent Builder is scheduled to shut down.

See Migrate from Agent Builder to continue with the Agents SDK or ChatGPT Workspace Agents.

2026-06-02: GPT Image model deprecations

Shutdown dateModel / systemRecommended replacement
Dec 1, 2026gpt-image-1-minigpt-image-2
Dec 1, 2026gpt-image-1.5gpt-image-2
Dec 1, 2026chatgpt-image-latestgpt-image-2

For implementation examples, see Generate Images with GPT Image.

2026-05-08: GPT-5.2 and GPT-5.3 chat snapshots

Shutdown dateModel / systemRecommended replacement
Aug 10, 2026gpt-5.2-chat-latestgpt-5.5
Aug 10, 2026gpt-5.3-chat-latestgpt-5.5

2026-05-07: Self-serve fine-tuning availability

Inference on fine-tuned models remains available until their base models are deprecated.

DateUpdate
May 7, 2026Creating fine-tuning jobs or training becomes unavailable to organizations that have not previously run fine-tuning.
July 2, 2026Creating fine-tuning jobs becomes unavailable to organizations that have not run inference on a fine-tuned model in the previous 60 days.
Jan 6, 2027Active existing customers can no longer create new fine-tuning jobs. Inference is disabled only when the underlying base model is deprecated.

2026-04-22: Legacy GPT model snapshots

Shutdown dateModel snapshotSubstitute model
2026-07-23computer-use-preview-2025-03-11, computer-use-previewgpt-5.4-mini
2026-07-23gpt-4o-mini-search-preview-2025-03-11gpt-5.4-mini
2026-07-23gpt-4o-mini-tts-2025-03-20gpt-4o-mini-tts-2025-12-15
2026-07-23gpt-4o-search-preview-2025-03-11gpt-5.4-mini
2026-07-23gpt-5-chat-latestgpt-5.5
2026-07-23gpt-5-codexgpt-5.5
2026-07-23gpt-5.1-chat-latestgpt-5.5
2026-07-23gpt-5.1-codexgpt-5.5
2026-07-23gpt-5.1-codex-maxgpt-5.5
2026-07-23gpt-5.1-codex-minigpt-5.4-mini
2026-07-23gpt-audio-mini-2025-10-06gpt-audio-1.5
2026-07-23gpt-realtime-mini-2025-10-06gpt-realtime-mini
2026-07-23o3-deep-research-2025-06-26, o3-deep-researchgpt-5.5-pro
2026-07-23o4-mini-deep-research-2025-06-26, o4-mini-deep-researchgpt-5.5-pro
2026-07-23gpt-5.2-codexgpt-5.5
2026-10-23gpt-3.5-turbo-0125, gpt-3.5-turbo, gpt-3.5-turbo-completionsgpt-5.4-mini
2026-10-23gpt-4-0613, gpt-4, gpt-4-0613-completions, gpt-4-completionsgpt-5.5
2026-10-23gpt-4-1106-previewgpt-5.5
2026-10-23gpt-4-turbo, gpt-4-turbo-2024-04-09, gpt-4-turbo-completionsgpt-5.5
2026-10-23gpt-4.1-nano, gpt-4.1-nano-2025-04-14gpt-5.4-nano
2026-10-23gpt-4o-2024-05-13gpt-5.5
2026-10-23gpt-image-1gpt-image-2
2026-10-23o1-2024-12-17, o1gpt-5.5
2026-10-23o1-pro-2025-03-19, o1-progpt-5.5-pro
2026-10-23o3-mini-2025-01-31, o3-minigpt-5.5
2026-10-23ft-o4-mini-2025-04-16gpt-5.4-mini
2026-10-23o4-mini-2025-04-16, o4-minigpt-5.4-mini

The following fine-tuned versions are also being removed:

Shutdown dateModel snapshotRecommended replacement base model
2026-10-23ft-gpt-3.5-turbogpt-5.4-mini
2026-10-23ft-gpt-4gpt-5.5
2026-10-23ft-gpt-4.1-nano-2025-04-14gpt-5.4-nano
2026-10-23ft-babbage-002gpt-5.4-mini
2026-10-23ft-davinci-002gpt-5.4-mini

2026-03-24: Sora 2 models and Videos API

Shutdown dateModel / systemRecommended replacement
2026-09-24Videos API
2026-09-24sora-2
2026-09-24sora-2-pro
2026-09-24sora-2-2025-10-06
2026-09-24sora-2-2025-12-08
2026-09-24sora-2-pro-2025-10-06

2025-11-14: DALL·E model snapshots

Shutdown dateModel / systemRecommended replacement
2026-05-12dall-e-2gpt-image-2, gpt-image-1, or gpt-image-1-mini
2026-05-12dall-e-3gpt-image-2, gpt-image-1, or gpt-image-1-mini

2025-09-26: Legacy GPT model snapshots

Shutdown dateModel / systemRecommended replacement
2026-03-26gpt-4-0314gpt-5 or gpt-4.1*
2026-03-26gpt-4-1106-previewgpt-5 or gpt-4.1*
2026-03-26gpt-4-0125-preview, gpt-4-turbo-preview, gpt-4-turbo-preview-completionsgpt-5 or gpt-4.1*
2026-09-28gpt-3.5-turbo-instructgpt-5.4-mini or gpt-5-mini
2026-09-28babbage-002gpt-5.4-mini or gpt-5-mini
2026-09-28davinci-002gpt-5.4-mini or gpt-5-mini
2026-09-28gpt-3.5-turbo-1106gpt-5.4-mini or gpt-5-mini

*For tasks that are especially latency-sensitive and do not require reasoning.

2025-09-15: Realtime API Beta

The Realtime API Beta was removed from the OpenAI API on May 12, 2026. See the beta-to-GA migration guide.

Shutdown dateModel / systemRecommended replacement
2026-05-12OpenAI-Beta: realtime=v1Realtime API

2025-08-20: Assistants API

The Assistants API is scheduled to shut down one year after affected developers were notified on August 26, 2025. See the Assistants-to-Conversations migration guide.

Shutdown dateModel / systemRecommended replacement
2026-08-26Assistants APIResponses API and Conversations API

2025-09-15: GPT-4o realtime and audio preview models

Shutdown dateModel / systemRecommended replacement
2026-05-07gpt-4o-realtime-previewgpt-realtime-1.5
2026-05-07gpt-4o-realtime-preview-2025-06-03gpt-realtime-1.5
2026-05-07gpt-4o-realtime-preview-2024-12-17gpt-realtime-1.5
2026-05-07gpt-4o-mini-realtime-previewgpt-realtime-mini
2026-05-07gpt-4o-audio-previewgpt-audio-1.5
2026-05-07gpt-4o-mini-audio-previewgpt-audio-mini

Past OpenAI Deprecations

2025-11-18: chatgpt-4o-latest

Shutdown dateModel / systemRecommended replacement
2026-02-17chatgpt-4o-latestgpt-5.1-chat-latest

2025-11-17: codex-mini-latest

OpenAI also ended support for the legacy local shell tool that was available only with this model.

Shutdown dateModel / systemRecommended replacement
2026-02-12codex-mini-latestgpt-5-codex-mini

2025-06-10: Early GPT-4o realtime and audio previews

Shutdown dateModel / systemRecommended replacement
2025-10-10gpt-4o-realtime-preview-2024-10-01gpt-realtime-1.5
2025-10-10gpt-4o-audio-preview-2024-10-01gpt-audio-1.5

2025-04-28: Text moderation

Shutdown dateModel / systemRecommended replacement
2025-10-27text-moderation-007omni-moderation
2025-10-27text-moderation-stableomni-moderation
2025-10-27text-moderation-latestomni-moderation

2025-04-28: o1-preview and o1-mini

Shutdown dateModel / systemRecommended replacement
2025-07-28o1-previewo3
2025-10-27o1-minio4-mini

2025-04-14: GPT-4.5 Preview

Shutdown dateModel / systemRecommended replacement
2025-07-14gpt-4.5-previewgpt-4.1

2024-10-02: Assistants API beta v1

See the Assistants API migration guide.

Shutdown dateModel / systemRecommended replacement
2024-12-18OpenAI-Beta: assistants=v1OpenAI-Beta: assistants=v2

2024-08-29: Fine-tuning training on babbage-002 and davinci-002

Fine-tuned models already created from these base models were not affected by this training deprecation.

Shutdown dateModel / systemRecommended replacement
2024-10-28New fine-tuning training on babbage-002gpt-4o-mini
2024-10-28New fine-tuning training on davinci-002gpt-4o-mini

2024-06-06: GPT-4 32K and Vision Preview models

Shutdown dateDeprecated modelDeprecated model priceRecommended replacement
2025-06-06gpt-4-32k$60.00 / 1M input tokens + $120 / 1M output tokensgpt-4o
2025-06-06gpt-4-32k-0613$60.00 / 1M input tokens + $120 / 1M output tokensgpt-4o
2025-06-06gpt-4-32k-0314$60.00 / 1M input tokens + $120 / 1M output tokensgpt-4o
2024-12-06gpt-4-vision-preview$10.00 / 1M input tokens + $30 / 1M output tokensgpt-4o
2024-12-06gpt-4-1106-vision-preview$10.00 / 1M input tokens + $30 / 1M output tokensgpt-4o

2023-11-06: Chat model updates

Fine-tuned models created from these bases were not affected, but new fine-tuned versions could no longer be created.

Shutdown dateDeprecated modelDeprecated model priceRecommended replacement
2024-09-13gpt-3.5-turbo-0613$1.50 / 1M input tokens + $2.00 / 1M output tokensgpt-3.5-turbo
2024-09-13gpt-3.5-turbo-16k-0613$3.00 / 1M input tokens + $4.00 / 1M output tokensgpt-3.5-turbo

2023-08-22: Fine-tunes endpoint

Models fine-tuned through /v1/fine-tunes also became inaccessible after shutdown.

Shutdown dateSystemRecommended replacement
2024-01-04/v1/fine-tunes/v1/fine_tuning/jobs

2023-07-06: GPT and embeddings

InstructGPT models

Shutdown dateDeprecated modelDeprecated model priceRecommended replacement
2024-01-04text-ada-001$0.40 / 1M tokensgpt-3.5-turbo-instruct
2024-01-04text-babbage-001$0.50 / 1M tokensgpt-3.5-turbo-instruct
2024-01-04text-curie-001$2.00 / 1M tokensgpt-3.5-turbo-instruct
2024-01-04text-davinci-001$20.00 / 1M tokensgpt-3.5-turbo-instruct
2024-01-04text-davinci-002$20.00 / 1M tokensgpt-3.5-turbo-instruct
2024-01-04text-davinci-003$20.00 / 1M tokensgpt-3.5-turbo-instruct

Base GPT models

Shutdown dateDeprecated modelDeprecated model priceRecommended replacement
2024-01-04ada$0.40 / 1M tokensbabbage-002
2024-01-04babbage$0.50 / 1M tokensbabbage-002
2024-01-04curie$2.00 / 1M tokensdavinci-002
2024-01-04davinci$20.00 / 1M tokensdavinci-002
2024-01-04code-davinci-002gpt-3.5-turbo-instruct

Edit models and endpoint

Shutdown dateModel / systemRecommended replacement
2024-01-04text-davinci-edit-001gpt-4o
2024-01-04code-davinci-edit-001gpt-4o
2024-01-04/v1/edits/v1/chat/completions

Fine-tuning GPT models

Shutdown dateDeprecated modelTraining priceUsage priceRecommended replacement
2024-01-04ada$0.40 / 1M tokens$1.60 / 1M tokensbabbage-002
2024-01-04babbage$0.60 / 1M tokens$2.40 / 1M tokensbabbage-002
2024-01-04curie$3.00 / 1M tokens$12.00 / 1M tokensdavinci-002
2024-01-04davinci$30.00 / 1M tokens$120.00 / 1K tokensdavinci-002, gpt-3.5-turbo, or gpt-4o

First-generation text embedding models

Shutdown dateDeprecated modelDeprecated model priceRecommended replacement
2024-01-04text-similarity-ada-001$4.00 / 1M tokenstext-embedding-3-small
2024-01-04text-search-ada-doc-001$4.00 / 1M tokenstext-embedding-3-small
2024-01-04text-search-ada-query-001$4.00 / 1M tokenstext-embedding-3-small
2024-01-04code-search-ada-code-001$4.00 / 1M tokenstext-embedding-3-small
2024-01-04code-search-ada-text-001$4.00 / 1M tokenstext-embedding-3-small
2024-01-04text-similarity-babbage-001$5.00 / 1M tokenstext-embedding-3-small
2024-01-04text-search-babbage-doc-001$5.00 / 1M tokenstext-embedding-3-small
2024-01-04text-search-babbage-query-001$5.00 / 1M tokenstext-embedding-3-small
2024-01-04code-search-babbage-code-001$5.00 / 1M tokenstext-embedding-3-small
2024-01-04code-search-babbage-text-001$5.00 / 1M tokenstext-embedding-3-small
2024-01-04text-similarity-curie-001$20.00 / 1M tokenstext-embedding-3-small
2024-01-04text-search-curie-doc-001$20.00 / 1M tokenstext-embedding-3-small
2024-01-04text-search-curie-query-001$20.00 / 1M tokenstext-embedding-3-small
2024-01-04text-similarity-davinci-001$200.00 / 1M tokenstext-embedding-3-small
2024-01-04text-search-davinci-doc-001$200.00 / 1M tokenstext-embedding-3-small
2024-01-04text-search-davinci-query-001$200.00 / 1M tokenstext-embedding-3-small

2023-06-13: Updated chat models

Shutdown dateLegacy modelLegacy model priceRecommended replacement
At earliest 2024-06-13gpt-4-0314$30.00 / 1M input tokens + $60.00 / 1M output tokensgpt-4o
Shutdown dateDeprecated modelDeprecated model priceRecommended replacement
2024-09-13gpt-3.5-turbo-0301$15.00 / 1M input tokens + $20.00 / 1M output tokensgpt-3.5-turbo
2025-06-06gpt-4-32k-0314$60.00 / 1M input tokens + $120.00 / 1M output tokensgpt-4o

2023-03-20: Codex models

Shutdown dateDeprecated modelRecommended replacement
2023-03-23code-davinci-002gpt-4o
2023-03-23code-davinci-001gpt-4o
2023-03-23code-cushman-002gpt-4o
2023-03-23code-cushman-001gpt-4o

2022-06-03: Legacy endpoints

Shutdown dateSystemRecommended replacement
2022-12-03/v1/engines/v1/models
2022-12-03/v1/searchSearch transition guide
2022-12-03/v1/classificationsClassifications transition guide
2022-12-03/v1/answersAnswers transition guide

Currently Deprecated Models

Google Gemma 3 Models (Deprecated: 2026-07-11)

The following Google Gemma 3 open models have been deprecated and are no longer available through AvalAI:

  • gemma-3n-e4b-it - Gemma 3n E4B instruction-tuned model
  • gemma-3n-e2b-it - Gemma 3n E2B instruction-tuned model
  • gemma-3-27b-it - Gemma 3 27B instruction-tuned model
  • gemma-3-12b-it - Gemma 3 12B instruction-tuned model
  • gemma-3-4b-it - Gemma 3 4B instruction-tuned model
  • gemma-3-1b-it - Gemma 3 1B instruction-tuned model

Migration Path: Use currently supported Gemini models available through AvalAI, such as gemini-2.5-flash or gemini-2.5-flash-lite, for chat and instruction-following workloads.

Z.AI GLM Legacy Models (Deprecated: 2026-06-27)

Z.AI has deprecated the following legacy GLM models in favor of the latest flagship glm-5.2 model:

  • glm-5v-turbo - Legacy GLM-5v Turbo vision model
  • glm-5-turbo - Legacy GLM-5 Turbo model
  • glm-5 - Legacy GLM-5 model
  • glm-4.7-flashx - Legacy GLM-4.7 FlashX model
  • glm-4.7-flash - Legacy GLM-4.7 Flash model
  • glm-4.7 - Legacy GLM-4.7 model
  • glm-4.6 - Legacy GLM-4.6 model

Migration Path: Please migrate to the latest supported Z.AI GLM model:

  • glm-5.2 - Latest flagship GLM model for complex software engineering, long-context reasoning, and stable agentic workflows, with a 1M-token context window and industry-leading coding performance

For details on the replacement model, see the Z.AI Models page.

MiniMax Legacy Models (Deprecated: 2026-06-27)

MiniMax has deprecated the following legacy models in favor of the latest flagship minimax-m3 model:

  • minimax-m2.1 - Legacy MiniMax M2.1 reasoning model
  • minimax-m2.1-lightning - Legacy MiniMax M2.1 Lightning model
  • minimax-m2 - Legacy MiniMax M2 model

Migration Path: Please migrate to the latest supported MiniMax model:

  • minimax-m3 - Latest flagship MiniMax model combining frontier coding capability, an ultra-long 1M-token context window, and native multimodality, built on the MiniMax Sparse Attention (MSA) architecture with toggleable thinking and autonomous long-horizon task support

For details on the replacement model, see the MiniMax Models page.

Google Programmable Search Engine (google_pse-search) (Deprecated: April 2026)

Google has deprecated the Programmable Search Engine (PSE) v1/search model, previously available as google_pse-search. This service has been discontinued by Google in favor of their AI-powered search capabilities built into Gemini models.

Deprecated Model:

  • google_pse-search - Google Programmable Search Engine v1/search model

Migration Path: You can either use the built-in search tool in the Gemini models, or switch to another v1/search provider:

  • Gemini models with google_search tool - Built-in AI-powered search in Gemini 3 Flash, Gemini 3.1 Pro, and Gemini 2.5 Pro models. See the Web Search Capabilities Guide for implementation details.
  • Other v1/search providers - Switch to an alternative search provider such as serper-search (lowest-cost Google-powered search). See the Search API reference for the full list of supported v1/search providers.

Mistral Codestral Models (Deprecated: 2026-06-03)

The following Mistral Codestral model has been deprecated:

  • codestral-2501 - Legacy Codestral code generation model (January 2025 snapshot)

Migration Path: Use the latest supported Mistral coding models:

  • codestral-latest - Latest Codestral model for code generation and completion

Cloudflare Workers AI Models (Deprecated: 2026-06-03)

The following Cloudflare Workers AI models have been deprecated and are no longer available:

  • cf.qwen3-embedding-0.6b - Qwen3 0.6B embedding model
  • cf.meta-llama-3-8b-instruct - Meta Llama 3 8B Instruct model
  • cf.llama-3.1-8b-instruct-awq - Llama 3.1 8B Instruct (AWQ quantized)
  • cf.llama-3.1-8b-instruct-fp8 - Llama 3.1 8B Instruct (FP8 quantized)
  • cf.llama-3.1-8b-instruct - Llama 3.1 8B Instruct model
  • cf.llama-3-8b-instruct-awq - Llama 3 8B Instruct (AWQ quantized)
  • cf.llama-3-8b-instruct - Llama 3 8B Instruct model
  • cf.gemma-7b-it-lora - Gemma 7B Instruct (LoRA)
  • cf.gemma-2b-it-lora - Gemma 2B Instruct (LoRA)
  • cf.gemma-7b-it - Gemma 7B Instruct model
  • cf.llama-3.1-70b-instruct - Llama 3.1 70B Instruct model

Migration Path: Use currently supported models for chat, embeddings, and code generation available through AvalAI.

Google Gemini Preview & Experimental Models (Deprecated: 2025-11-18)

Google has deprecated the following preview and experimental Gemini models. Users should migrate to currently supported Gemini models, including gemini-3.1-pro-preview for advanced reasoning tasks.

Deprecated Models:

  • gemini-2.0-flash-exp
  • gemini-2.0-flash-lite-preview
  • gemini-2.0-flash-lite-preview-02-05
  • gemini-2.0-flash-thinking-exp
  • gemini-2.0-flash-thinking-exp-01-21
  • gemini-2.0-flash-thinking-exp-1219
  • gemini-2.5-flash-lite-preview-06-17
  • gemini-2.5-flash-preview-05-20
  • gemini-2.5-pro-preview-06-05
  • gemini-2.5-pro-preview-03-25
  • gemini-2.5-pro-preview-05-06
  • gemini-3-pro-preview

Migration Path: Use currently supported models:

  • gemini-3.1-pro-preview - Latest Gemini Pro Preview model for advanced reasoning tasks
  • gemini-2.5-pro - Stable Gemini Pro model
  • gemini-2.5-flash - Stable Gemini Flash model
  • gemini-2.5-flash-preview-09-2025 - Gemini Flash preview (September 2025)
  • gemini-2.5-flash-lite - Gemini Flash Lite model
  • gemini-2.5-flash-lite-preview-09-2025 - Gemini Flash Lite preview (September 2025)

For more information, see our announcement.

Google Gemini Image & Embedding Models (Deprecated: 2025-10-28)

The following Google Gemini image and embedding models have been deprecated:

Image Generation Models

  • gemini-2.5-flash-image-preview - Preview version of Gemini 2.5 Flash image generation model

Migration Path: Use currently supported models:

  • gemini-2.5-flash-image - Stable version of Gemini 2.5 Flash image generation

Embedding Models

  • embedding-001 - Legacy embedding model
  • embedding-gecko-001 - Legacy Gecko embedding model
  • gemini-embedding-exp-03-07 - March 7 experimental embedding model
  • gemini-embedding-exp - Experimental embedding model

Migration Path: Use currently supported embedding models:

  • text-embedding-004 - Latest text embedding model
  • text-multilingual-embedding-002 - Multilingual embedding model

BytePlus Seedream Models (Deprecated: 2026-05-06)

The following BytePlus Seedream image generation models have been deprecated:

  • seedream-4-0-250828 - Legacy Seedream 4.0 image generation and editing model
  • seedream-4-5-251128 - Legacy Seedream 4.5 image generation and editing model

Migration Path: Use the latest supported Seedream model:

  • seedream-5-0-260128 - Latest Seedream 5.0 image generation and editing model with Chain of Thought reasoning, improved prompt optimization, MJ-style aesthetics, and enhanced high-resolution generation

For implementation examples, see Generate Images with Seedream.

OpenAI Models (Deprecated: 2025-10-28)

The following OpenAI models have been deprecated:

Image Generation Models

  • dall-e-3 - Legacy DALL·E 3 image generation model

Migration Path: Use the latest OpenAI image model:

  • gpt-image-2 - Latest OpenAI image generation and editing model with improved prompt adherence, visual fidelity, multilingual text rendering, and support for both v1/images/generations and v1/images/edits

For implementation examples, see Generate Images with GPT Image.

GPT-3.5 Models

  • gpt-3.5-turbo - Legacy GPT-3.5 Turbo model

Migration Path: Use currently supported models:

  • gpt-5-nano - More efficient and capable replacement
  • gpt-5-mini - Enhanced performance model

GPT-4 Legacy Models

  • gpt-4 - Original GPT-4 model
  • gpt-4-0125-preview - January 25, 2024 preview version
  • gpt-4-1106-preview - November 6, 2023 preview version
  • gpt-4-turbo - Legacy GPT-4 Turbo model
  • gpt-4-turbo-2024-04-09 - April 9, 2024 specific version

Migration Path: Use currently supported models:

  • gpt-4.1 - Latest GPT-4 series model
  • gpt-5-chat - Next generation model
  • gpt-5-pro - Professional tier model

Reasoning Models

  • o1-preview - Preview version of O1 reasoning model
  • gpt-4.5-preview - Preview version of GPT-4.5

Migration Path: Use currently supported models:

  • o3 - Stable version of O1 reasoning model
  • o4-mini - Efficient reasoning model
  • gpt-5-chat - Latest Chat model

Anthropic Claude Models (Deprecated: 2025-10-28)

The following Anthropic Claude models have been deprecated:

  • anthropic.claude-3-opus-20240229-v1:0 - Claude 3 Opus (February 29, 2024 version)
  • anthropic.claude-3-haiku-20240307-v1:0 - Claude 3 Haiku (March 7, 2024 version)
  • anthropic.claude-3-sonnet-20240229-v1:0 - Claude 3 Sonnet (February 29, 2024 version)
  • anthropic.claude-3-5-haiku-20241022-v1:0 - Claude 3.5 Haiku (October 22, 2024 version)
  • anthropic.claude-3-5-sonnet-20240620-v1:0 - Claude 3.5 Sonnet (June 20, 2024 version)
  • anthropic.claude-3-7-sonnet-20250219-v1:0 - Claude 3.7 Sonnet (February 19, 2025 version)

Migration Path: Use currently supported models:

  • claude-opus-4-8 - Latest Claude Opus model
  • claude-sonnet-4-6 - Latest Claude Sonnet model
  • claude-haiku-4-5 - Efficient alternative

Stability AI Models (Deprecated: 2025-10-28)

The following Stability AI models have been deprecated:

  • stability.sd3-large-v1:0 - Stable Diffusion 3 Large v1.0

Migration Path: Use currently supported models:

  • stability.sd3-5-large-v1:0 - Latest Stable Diffusion 3.5 model
  • stability.stable-image-ultra-v1:1 - Alternative image generation model

Google Gemini Models (Deprecated: 2025-09-27)

The following Google Gemini models have been officially deprecated by Google's Gemini API:

Core Gemini Models

  • gemini-pro - Legacy general-purpose model
  • gemini-1.5-pro - Previous generation pro model
  • gemini-1.5-pro-002 - Specific version variant
  • gemini-1.5-pro-001 - Specific version variant
  • gemini-1.5-pro-latest - Latest pointer (deprecated)

Experimental Variants

  • gemini-1.5-pro-exp-0801 - August 2024 experimental version
  • gemini-1.5-pro-exp-0827 - August 27, 2024 experimental version
  • gemini-2.0-flash-exp - Flash 2.0 experimental model
  • gemini-exp-1114 - November 14 experimental model
  • gemini-exp-1206 - December 6 experimental model

Flash Models

  • gemini-1.5-flash-latest - Latest flash model pointer
  • gemini-1.5-flash-8b - 8 billion parameter flash model
  • gemini-1.5-flash-8b-exp-0924 - September 24 experimental flash variant
  • gemini-1.5-flash-exp-0827 - August 27 experimental flash variant
  • gemini-1.5-flash-8b-exp-0827 - August 27 experimental 8B flash variant

Migration Path: Use currently supported models:

  • gemini-2.0-flash - Generation 2.0 flash model
  • gemini-2.0-flash-lite - Generation 2.0 flash lite model
  • gemini-2.5-pro - Generation 2.5 pro model
  • gemini-2.5-flash - Generation 2.5 flash model
  • gemini-2.5-flash-lite - Generation 2.5 flash lite model

Announcement: Google Gemini 1.5 Series Models Deprecation Notice

Migration Support

If you need assistance migrating from deprecated models:

  1. Review Current Usage: Identify which deprecated models you're currently using
  2. Select Alternatives: Choose appropriate replacement models based on your use case
  3. Update Implementation: Modify your code to use the new model names
  4. Test Thoroughly: Ensure your applications work correctly with the replacement models
  5. Monitor Performance: Verify that the new models meet your requirements

Support Resources

Historical Deprecations

This section will be updated as models are deprecated over time.


Last Updated: 2026-07-24

For the most current information about model availability, please refer to our Models Documentation or contact our support team.