October was a particularly significant month for ModelCosts, with OpenAI's large-scale model retirement programme dominating the changes. There were 15 formal retirements and 16 moves to legacy status, alongside 17 models disappearing from provider offerings. Anthropic and OpenAI both reduced cached-input pricing for Claude Sonnet 5.5, while five new models entered the market, including Gemini 4 Argon and Nemotron-3.5 Super. The month's changes point to a broader shift towards newer model generations, with providers consolidating older model lineups and setting retirement dates for existing offerings.
In October 2026, we recorded 3 price changes, 5 new model releases, 1 model newly listed on another provider, 17 models providers stopped offering, 16 moves to legacy and 15 model retirements.
Price changes
- 8th Oct: Anthropic cut the price of Claude Sonnet 5.5: Cached input $0.20 → $0.10 per 1M tokens (−50%).
- 13th Oct: Perplexity followed Anthropic to cut its Claude Sonnet 5.5: Cached input $0.20 → $0.10 per 1M tokens (−50%).
Batch, Priority and other tier price changes
- 8th Oct: Anthropic cut the Batch price of Claude Sonnet 5.5: Cached input $0.10 → $0.05 per 1M tokens (−50%).
New models
- 7th Oct: Anthropic released Claude Haiku 5.5.
- 10th Oct: OpenAI released GPT Rosalind Discovery.
- 13th Oct: Google released Gemini 4 Argon.
- 15th Oct: NVIDIA released Nemotron-3.5 Super-120b-a12b.
- 22nd Oct: Z.ai released GLM 5.3 FlashX Legal.
Now on more providers
- 15th Oct: Claude Haiku 5.5 is now listed on Perplexity.
No longer on a provider
- 16th Oct: Perplexity stopped offering NVIDIA's Nemotron-3 Ultra-550b-a55b, and no tracked provider now offers it.
- 17th Oct: Perplexity stopped offering 16 models:
- Anthropic's Claude Haiku 4.5
- Anthropic's Claude Opus 4.5
- Anthropic's Claude Opus 4.6
- Anthropic's Claude Opus 4.7
- Anthropic's Claude Opus 4.8
- Anthropic's Claude Sonnet 4.5
- Anthropic's Claude Sonnet 4.6
- Google's Gemini 3 Flash Preview, which no tracked provider now offers
- Google's Gemini 3.1 Flash Lite, which no tracked provider now offers
- Google's Gemini 3.1 Pro Preview, which no tracked provider now offers
- Google's Gemini 3.5 Flash, which no tracked provider now offers
- Google's Gemini 3.5 Flash Lite, which no tracked provider now offers
- xAI's Grok-4.20 Multi-Agent, which no tracked provider now offers
- xAI's Grok-4.20 Non-Reasoning, which no tracked provider now offers
- xAI's Grok-4.20 Reasoning, which no tracked provider now offers
- xAI's Grok-4.3, which no tracked provider now offers
Moved to legacy
- 4th Oct: Anthropic moved Claude Sonnet 4.5 to legacy, retiring it on 30th Nov 2026, with Claude Sonnet 5.5 as the suggested replacement.
- 4th Oct: OpenAI moved GPT-4o Mini TTS to legacy, retiring it on 6th Jan 2027, with GPT Realtime 2.1 Mini as the suggested replacement.
- 4th Oct: OpenAI moved GPT-4o Mini Transcribe to legacy, retiring it on 26th Feb 2027, with GPT Transcribe as the suggested replacement.
- 4th Oct: OpenAI moved GPT-4o Transcribe to legacy, retiring it on 26th Feb 2027, with GPT Transcribe as the suggested replacement.
- 4th Oct: OpenAI moved GPT-4o Transcribe Diarize to legacy, retiring it on 26th Feb 2027, with GPT Transcribe as the suggested replacement.
- 4th Oct: OpenAI moved GPT-5 to legacy, retiring it on 11th Dec 2026, with GPT-5.6 Sol as the suggested replacement.
- 4th Oct: OpenAI moved GPT-5 Mini to legacy, retiring it on 11th Dec 2026, with GPT-5.6 Terra as the suggested replacement.
- 4th Oct: OpenAI moved GPT-5 Nano to legacy, retiring it on 11th Dec 2026, with GPT-5.6 Luna as the suggested replacement.
- 4th Oct: OpenAI moved GPT-5 Pro to legacy, retiring it on 11th Dec 2026, with GPT-5.6 Sol as the suggested replacement.
- 4th Oct: OpenAI moved GPT-5.1 to legacy, retiring it on 1st Apr 2027, with GPT-6 Sol as the suggested replacement.
- 4th Oct: OpenAI moved GPT-5.3 Codex to legacy, retiring it on 1st Apr 2027, with GPT-6 Sol as the suggested replacement.
- 4th Oct: OpenAI moved GPT-5.4 Nano to legacy, retiring it on 1st Apr 2027, with GPT-6 Luna as the suggested replacement.
- 4th Oct: OpenAI moved TTS-1 to legacy, retiring it on 6th Jan 2027, with GPT Realtime 2.1 Mini as the suggested replacement.
- 4th Oct: OpenAI moved TTS-1 HD to legacy, retiring it on 6th Jan 2027, with GPT Realtime 2.1 Mini as the suggested replacement.
- 4th Oct: OpenAI moved o3 to legacy, retiring it on 11th Dec 2026, with GPT-5.6 Sol as the suggested replacement.
- 4th Oct: OpenAI moved o3 Pro to legacy, retiring it on 11th Dec 2026, with GPT-5.6 Sol as the suggested replacement.
Retired
- 23rd Oct: OpenAI retired GPT Image 1, with GPT Image 2 as the suggested replacement.
- 23rd Oct: OpenAI retired GPT-3.5 Turbo, with GPT-5.6 Terra as the suggested replacement.
- 23rd Oct: OpenAI retired GPT-3.5 Turbo 0125, with GPT-5.6 Terra as the suggested replacement.
- 23rd Oct: OpenAI retired GPT-4 0613, with GPT-5.6 Sol as the suggested replacement.
- 23rd Oct: OpenAI retired GPT-4 1106 Preview, with GPT-5.6 Sol as the suggested replacement.
- 23rd Oct: OpenAI retired GPT-4 Turbo 2024-04-09, with GPT-5.6 Sol as the suggested replacement.
- 23rd Oct: OpenAI retired GPT-4.1 Nano, with GPT-5.6 Luna as the suggested replacement.
- 23rd Oct: OpenAI retired GPT-4.1 Nano 2025-04-14, with GPT-5.6 Luna as the suggested replacement.
- 23rd Oct: OpenAI retired GPT-4o 2024-05-13, with GPT-5.6 Sol as the suggested replacement.
- 23rd Oct: OpenAI retired o1, with GPT-5.6 Sol as the suggested replacement.
- 23rd Oct: OpenAI retired o1 Pro, with GPT-5.6 Sol as the suggested replacement.
- 23rd Oct: OpenAI retired o3 Mini, with GPT-5.6 Sol as the suggested replacement.
- 23rd Oct: OpenAI retired o4 Mini, with GPT-5.6 Terra as the suggested replacement.
- 23rd Oct: OpenAI retired o4 Mini 2025-04-16, with GPT-5.6 Terra as the suggested replacement.
- 23rd Oct: OpenAI retired o4 Mini 2025-04-16 with data sharing, with GPT-5.6 Terra as the suggested replacement.
Track every change, as it happens
This report captures the notable changes recorded during October 2026. AI model pricing, availability and lifecycle status continue to evolve throughout the year.
Follow the changes page for ongoing updates, or visit our methodology page to understand how we collect, verify and track model pricing.