Skip to main content

​
Changelog

​

​
Deprecating legacy models:

We’re retiring several models from our catalog to ensure you always have access to the latest generation of highest quality and performant models. These models will stop accepting requests starting August 14, 2026.
  • Deprecated: 2026-08-11 — deprecated models continue to serve requests, but responses carry a Deprecation header and a Link header pointing to their replacement.
  • Sunset: 2026-08-14 — after this date, deprecated models stop serving inference requests entirely.

​
Migration Guide

  • Claude Opus 4.1/4.5/4.6/4.7/4.8 — migrate to Claude Opus 5
  • Grok 4.5, Qwen3.6 Max Preview, Qwen3.7 Max, Sakana Fugu Ultra — migrate to Claude Opus 5
  • Poolside Laguna S 2.1, Meta Muse Spark 1.1 — migrate to Claude Opus 5
  • Gemini 3 Flash, 3.1 Flash Lite, 3.5 Flash, 3.5 Flash Lite, 3.6 Flash — migrate to Claude Sonnet 5
  • Claude Sonnet 4.5, 4.6, Qwen 3.6 Plus, 3.7 Plus — migrate to Claude Sonnet 5
  • Mistral Medium, Mistral Medium 3.5, Mistral Devstral 2, Magistral Medium — migrate to Claude Sonnet 5
  • Qwen 3.6 Flash, Ministral 3B, 14B, Mistral Devstral Small 2, Inkling Small — migrate to Claude Haiku 4.5
  • GPT-4.1, 4.1 mini, 4.1 nano, 4o, 4o mini — migrate to GPT-5.5
  • GPT-5 mini, 5 nano, 5.1, 5.3 Codex, 5.4, 5.4 mini, 5.4 nano — migrate to GPT-5.5
  • Llama 3.2 1B, 3.2 3B, 3.2 3B Instruct, 3.3 70B Instruct — migrate to Nemotron 3.5 Nano
  • Qwen2.5-Coder 0.5B, Qwen3 32B, 8B, 4B Base, 4B Instruct, 1.7B Base, Qwen3.5 9B, Qwen3.6 27B, Mistral 7B Instruct v0.3, Nemotron 3 Nano — migrate to Nemotron 3.5 Nano
  • Gemma 3 4B (Pretrained), Gemma 4 12B IT, 31B IT, E2B IT, E4B IT, SmolLM3 3B Base — migrate to Nemotron 3.5 Nano
  • DeepSeek V3 0324, V4 Pro, GPT-OSS 120B, 20B, LFM2 24B A2B — migrate to DeepSeek V4 Flash
  • Mistral Codestral, Magistral Small, Ministral 8B, Mistral Nemo, Pixtral 12B — migrate to DeepSeek V4 Flash
  • GLM 5.1, MiniMax M2.7, M3, MiMo V2.5, V2.5 Pro — migrate to GLM 5.2
  • Mistral Small 4, Qwen3.6 35B A3B, Qwen3 235B A22B Instruct, Nemotron 3 Super, Ultra, DiffusionGemma 26B-A4B IT — migrate to GLM 5.2
  • BGE-M3, Qwen3 Embedding 4B, 8B — migrate to text-embedding-3-large
​

​
Zero Data Retention (ZDR) visibility on the Models page

The Models page now shows a ZDR badge on every model that has at least one Zero Data Retention–compliant provider route, so you can filter for ZDR support without checking each model’s routing individually.See Trust & Safety for what ZDR means on Pioneer and how it’s implemented per provider.
​

​
New model: DeepSeek V4 Flash 0731

DeepSeek V4 Flash 0731 is now available in the Model Catalog.
​

​
New model: GLM 5.2

GLM 5.2 is now available in the Model Catalog.