Full Changelog

GPT-6 Sol released with a 1,050,000-token context window, 128K max output, and an April 20, 2026 knowledge cutoff

API pricing set at $2.00 per 1M input tokens and $10.00 per 1M output tokens, half of GPT-5.6 Sol; cached input $0.20 per 1M tokens; OpenAI says the cut is permanent rather than promotional

GPT-6 Luna released with a 1,050,000-token context window, 128K max output, and a May 18, 2026 knowledge cutoff; supports reasoning effort from none to max

API pricing set at $0.10 per 1M input tokens and $0.50 per 1M output tokens, half of GPT-5.6 Luna; cached input $0.01 per 1M tokens

Succeeded by GPT-6 Sol, which launched at half the price

Succeeded by GPT-6 Luna, which launched at half the price

Claude Opus 5.5 released with a 1M context window, up to 128K output tokens, always-on adaptive thinking, and a June 2026 knowledge cutoff

Pricing set at $4.00 per 1M input tokens and $20.00 per 1M output tokens, 20% below Opus 5; cache reads drop to $0.20 per 1M tokens

Breaking changes from Opus 5: thinking can no longer be disabled, forced tool use returns an error, thinking blocks are bound to the model and conversation, and the computer_20251124 tool was removed; default effort changed from high to medium

Anthropic announced Claude Sonnet 5.5 and Claude Haiku 5.5 will follow in the coming weeks

Moved to the legacy model list after the release of Claude Opus 5.5; remains available for compatibility and migration comparisons

Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS released to general availability with voice design, voice replication, and a Voices endpoint

Access to Gemini 2.5 models limited to users who have actively used them before; not deprecated and still served through the API

Access to Gemini 2.5 models limited to users who have actively used them before; not deprecated and still served through the API

Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking audio-to-audio models released for the Live API

DeepSeek V4.1 Flash released with a 1M context window, up to 384K output, native vision, and support for both thinking and non-thinking modes; call it via the deepseek-flash model name

API pricing reduced to $0.30 per 1M input tokens and $1.20 per 1M output tokens at peak hours ($0.15 and $0.60 off-peak); cache hits $0.006 per 1M tokens

Planned routing of V4 Pro requests to V4.1 Flash after September 14, 2026 was reversed; V4 Pro remains on the API with unchanged billing

V4 Flash and V4 Flash Vision Exp retired on September 10, 2026; the deepseek-v4-flash and deepseek-v4-flash-vision-exp names are temporarily routed to V4.1 Flash

GPT-6 Astra rolled out to paid ChatGPT users in a restricted version that declines certain cybersecurity prompts

GPT-6 Astra unveiled and released as a limited preview with a 1,050,000-token context window, 128K max output, and an April 30, 2026 knowledge cutoff; OpenAI says it meets the Critical cybersecurity threshold under its Preparedness Framework

API pricing set at $10.00 per 1M input tokens and $50.00 per 1M output tokens (2.5x GPT-5.6 Sol); cached input $1.00, Batch at half price, Fast mode at 2x; requests above about 272K input tokens billed at $20.00 input and $75.00 output

Gemini 3.8 Flash released to general availability in the Gemini API as the recommended model for new projects

API pricing set at $0.75 per 1M input tokens and $3.75 per 1M output tokens through December 31, 2026, rising to $1.50 and $7.50 from January 1, 2027

Claude Fable 5.1 released alongside the restricted-access Claude Mythos 5.1, with a 1M context window, 128K max output, always-on adaptive thinking, and a June 2026 knowledge cutoff

Pricing unchanged from Fable 5 at $10.00 per 1M input tokens and $50.00 per 1M output tokens; cache reads cut 75% from $1.00 to $0.25 per 1M tokens

Forced tool use behavior changed (breaking change); Anthropic recommends starting workloads on Opus and escalating to Fable 5.1

Introductory rate of $2.00 input and $10.00 output per 1M tokens made the standard price; the previously scheduled increase to $3.00 and $15.00 was dropped

Superseded by Claude Fable 5.1 and moved to the legacy model list

Claude Mythos 5.1 released as the trusted-access sibling of Fable 5.1 for vetted cyberdefense and life-science organizations

Agentic video understanding added for Gemini 3.7 Flash, 3.6 Flash, and 3.5 Flash-Lite, using up to 88% fewer tokens on long-form video

Mistral OCR 4.1 reached general availability

API and credit pricing cut by more than 20% to $4.00 per 1M input tokens and $20.00 per 1M output tokens for three months, through at least November 21, 2026

Experimental multimodal variant DeepSeek-V4-Flash-Vision-Exp released on the API

Gemini 3.7 Flash released to general availability with substantial improvements in software engineering, web development, and agentic workflows

Introductory API pricing set at $0.75 per 1M input tokens and $3.75 per 1M output tokens through December 31, 2026

DeepSeek V4 Pro reached general availability as build 0813 with 1.6T total / 49B active parameters, 1M context, up to 384K output, low/high/max thinking effort, and native Responses API support

Peak/off-peak API pricing introduced from August 16, 2026: $1.32 per 1M input tokens and $3.96 per 1M output tokens at peak hours, half that off-peak

GPT-5.6-Cyber variant launched for cyber-defense partners

DeepSeek V4 Flash official release (build 0731) entered public beta with open weights under the MIT license, re-post-trained on the same architecture with much stronger agent benchmarks

Price reduced 20% to $2.00 per 1M input tokens and $12.00 per 1M output tokens

Price reduced 80% to $0.20 per 1M input tokens and $1.20 per 1M output tokens

Priority processing renamed Fast mode across the API; service_tier accepts either priority or fast

Claude Opus 5 released as the successor to Opus 4.8 with a 1M context window, adaptive thinking with low/medium/high/xhigh/max effort, and Fast Mode output about 2.5x faster

Pricing set at $5.00 per 1M input tokens and $25.00 per 1M output tokens; cached input $0.50 per 1M tokens

Claude Opus 5 replaced Opus 4.8 as the default model on the Claude Max plan

Superseded by Claude Opus 5 as the default Max plan model and moved to the legacy model list

The deepseek-chat and deepseek-reasoner API names were fully retired on July 24, 2026; use deepseek-flash or deepseek-v4-pro instead

Legacy deepseek-chat and deepseek-reasoner API names, which had pointed to V4 Flash, were discontinued on July 24, 2026

Gemini 3.6 Flash released to general availability, addressing developer feedback on output verbosity

API pricing set at $0.75 per 1M input tokens and $3.75 per 1M output tokens through December 31, 2026, half the price of Gemini 3.5 Flash

The temperature, top_p, and top_k sampling parameters were deprecated for the latest Gemini models

Gemini 3.5 Flash-Lite released to general availability as a high-volume subagent option

API pricing set at $0.30 per 1M input tokens and $2.50 per 1M output tokens

Google confirmed Gemini 3.5 Pro is still not ready and said it has started pretraining Gemini 4 as its next flagship

Fable 5 became a standard feature of Anthropic's higher-tier subscription plans after several extensions of the launch promotion

Mistral OCR 4.1 (mistral-ocr-4-1) released; mistral-ocr-latest and mistral-ocr-4 now point to it, and confidence scores gained block-level granularity

GPT-5.6 Sol released to the public as OpenAI's flagship and strongest coding and cybersecurity model, introducing the Sol/Terra/Luna capability-tier naming

Launch pricing set at $5.00 per 1M input tokens and $30.00 per 1M output tokens; Fast mode on Cerebras at $12.50 input and $75.00 output

GPT-5.6 Terra released as the mid-tier GPT-5.6 model, replacing the former mini tier

Launch pricing set at $2.50 per 1M input tokens and $15.00 per 1M output tokens

GPT-5.6 Luna released as the fastest and lowest-cost GPT-5.6 model, replacing the former nano tier

Launch pricing set at $1.00 per 1M input tokens and $6.00 per 1M output tokens

Succeeded by the GPT-5.6 family (Sol, Terra, Luna) as OpenAI's frontier lineup

Claude Sonnet 5 released with a 1M context window, 128K max output, adaptive thinking, and a January 2026 knowledge cutoff

Introductory pricing set at $2.00 per 1M input tokens and $10.00 per 1M output tokens through August 31, 2026, with no long-context surcharge

Sonnet 5 uses the updated tokenizer shared with Opus 4.7+ and Fable 5, producing roughly 1.0x to 1.35x as many tokens as Sonnet 4.6 for the same text

Leanstral 1.5 (labs-leanstral-1-5) released with improved SFT mixture quality and extended long-context reasoning for Lean 4 proof engineering

The Leanstral 1.5 labs model is scheduled for retirement on September 30, 2026

GPT-5.6 Sol, Terra, and Luna made available as a limited preview to about 20 trusted partner organizations while under U.S. government review

Computer Use tool launched in public preview for Gemini 3.5 Flash with browser, mobile, and desktop environments

Mistral OCR 4 (mistral-ocr-4-0) released with the new include_blocks option returning paragraph-level bounding boxes and structural labels

Access to Fable 5 and Mythos 5 was temporarily restricted following a U.S. Department of Commerce export-control directive; Anthropic later confirmed access was restored

Claude Fable 5 released as a Mythos-class model for general use, alongside Claude Mythos 5 for Project Glasswing partners; 1M context window and up to 128K output tokens

Pricing set at $10.00 per 1M input tokens and $50.00 per 1M output tokens, less than half of Claude Mythos Preview

Claude Mythos 5 released as a trusted-access preview to a small number of Project Glasswing partners; identical to Fable 5 apart from safeguards

Trusted-access pricing matches Fable 5 at $10.00 per 1M input tokens and $50.00 per 1M output tokens

gemini-2.0-flash and gemini-2.0-flash-lite were shut down on June 1, 2026; Google recommends gemini-3.5-flash or gemini-3.1-flash-lite instead

ChatGPT and API update improved GPT-5.5 Instant style and removed Canvas from GPT-5.5 Instant and Thinking

Updated in ChatGPT and the API for more natural, better-paced responses with fewer overly long or bullet-heavy answers

OpenAI announced that o3 will be retired from ChatGPT on August 26, 2026 after a 90-day sunset period

Claude Opus 4.8 released as Anthropic's latest high-capability model with a 1M context window

Pricing remains $5.00 per 1M input tokens and $25.00 per 1M output tokens

DeepSeek scheduled deepseek-chat and deepseek-reasoner compatibility aliases for retirement on June 15, 2026

Gemini 3.5 Flash released for Gemini API and Vertex AI with updated multimodal and reasoning performance

API pricing listed at $1.50 per 1M input tokens and $9.00 per 1M output tokens

GPT-5.5 Instant rolled out as ChatGPT's default model for all users and as chat-latest in the API, replacing GPT-5.3 Instant

Mistral Medium 3.5 released as the latest Medium family model with 262K context

Pricing set at $1.50 per 1M input tokens and $7.50 per 1M output tokens

GPT-5.5 became available in the API with a 1M context window

API pricing set at $5.00 per 1M input tokens and $30.00 per 1M output tokens

GPT-5.5 Pro became available in the API for higher-accuracy work

API pricing set at $30.00 per 1M input tokens and $180.00 per 1M output tokens

DeepSeek V4 Pro released for API access with stronger reasoning, coding, and tool-use performance

Promotional API pricing listed at $0.50 per 1M input tokens and $1.50 per 1M output tokens

DeepSeek V4 Flash released alongside V4 Pro for faster and lower-cost API workloads

Pricing set at $0.14 per 1M input tokens and $0.28 per 1M output tokens

GPT-5.5 introduced for ChatGPT and Codex as OpenAI's frontier model for complex professional work

GPT-5.5 Pro introduced as the high-compute GPT-5.5 mode for harder questions and higher-accuracy work

Gemini Embedding 2 made generally available for text embedding workloads

Meta confirmed Llama 4 Behemoth was still in training, leaving Maverick and Scout as the current released Llama 4 models

Claude Opus 4.7 released with a 1M context window, stronger tool use, and improved agentic coding performance

Pricing set at $5.00 per 1M input tokens and $25.00 per 1M output tokens

Claude Mythos Preview introduced as a gated research preview through Anthropic's Project Glasswing security initiative

Preview participant pricing listed at $25.00 per 1M input tokens and $125.00 per 1M output tokens after included usage credits

Voxtral TTS released for speech generation and voice output use cases

Gemini 3.1 Flash Live introduced for realtime audio and multimodal sessions

Leanstral released for formal mathematics and Lean-assisted reasoning tasks

Rolled out in ChatGPT as Thinking access for Free and Go users and as a fallback for GPT-5.4 Thinking limits

Gemini 3.1 Flash-Lite released for cost-sensitive, low-latency Gemini workloads

Mistral Small 4 released as the latest cost-efficient Small family model with 256K context

Pricing set at $0.15 per 1M input tokens and $0.45 per 1M output tokens

GPT-5.4 Mini released in the API, Codex, and ChatGPT with 400K context

API pricing set at $0.75 per 1M input tokens and $4.50 per 1M output tokens

GPT-5.4 Nano released for latency-sensitive and high-volume workloads in the API

API pricing set at $0.20 per 1M input tokens and $1.25 per 1M output tokens

API pricing set at $3.00 per 1M input tokens and $12.00 per 1M output tokens

Claude Opus 4.6 launched with improved reasoning and coding

Claude Sonnet 4.6 released alongside Opus 4.6

Gemini 3.1 Pro released with 2M context and enhanced reasoning

Initial release of GPT-5.4 with 1M context window

DeepSeek V3.2 released with improved reasoning and efficiency

Improved instruction following and reduced hallucination rate

Claude Opus 4.5 released with improved creative writing

Claude Sonnet 4.5 launched with improved coding performance

Initial release of GPT-5.2 with 512K context

Gemini 3 Pro launched with breakthrough multimodal capabilities

Gemini 3 Flash released with significantly improved speed

Improved multilingual performance

Claude Haiku 4.5 released with improved speed and capability

Deprecated in favor of Claude Haiku 4.5

Initial release of GPT-5.1

Claude Opus 4 launched as most capable Claude model to date

Claude Sonnet 4 launched alongside Opus 4

DeepSeek V3.1 released with improved instruction following

Price reduced from $3.00/$12.00 to $2.00/$8.00

Deprecated in favor of Claude 4 Opus

Deprecated in favor of Gemini 2.5+ series

GPT-5 launched — significant improvement over GPT-4 series

Deprecated in favor of GPT-4o and GPT-5 series

Updated to gemini-2.5-pro-preview-06-05 with improved performance

R1-0528 released with significant improvements in math reasoning

Improved function calling reliability

Gemini 2.5 Flash released with thinking capabilities at Flash pricing

o4-mini released with enhanced STEM reasoning capabilities

o3 launched with state-of-the-art reasoning on ARC-AGI and math benchmarks

GPT-4.1 released with 1M context window, best-in-class coding

GPT-4.1 Mini released alongside GPT-4.1

Llama 4 Maverick released as open-source 400B MoE model

Llama 4 Scout launched with industry-leading 10M token context

Gemini 2.5 Pro launched with built-in thinking capability

Claude 3.7 Sonnet launched with extended thinking capability

All Gemini 1.0 models deprecated

R1 API pricing reduced significantly

Gemini 2.0 Flash GA with stable API

o3-mini released as affordable reasoning model

Mistral Small 25.01 update with better performance

V3 API generally available with 128K context

DeepSeek R1 released as open-source reasoning model, rivaling o1

Mistral Medium relaunched with improved architecture

Codestral 25.01 with 256K context and improved performance

Deprecated — migrate to GPT-4o Mini

DeepSeek V3 launched, trained for just $5.6M

Full o1 model released with vision support and 200K context

Gemini 2.0 Flash experimental released

Llama 3.3 70B released, matching 405B quality at lower cost

Deprecated — migrate to Llama 3.x series

gpt-4o-2024-11-20 with improved creative writing

mistral-large-2411 update with improved instruction following

Pixtral Large 124B released with native multimodal understanding

Updated Claude 3.5 Sonnet with computer use capability

Context window expanded to 2M tokens

Mistral Small 24B released

o1-preview launched as first reasoning model

o1-mini launched alongside o1-preview

Quantized and distilled variants released by community

gpt-4o-2024-08-06 update with structured outputs

Mistral Large 2 (123B) released with 128K context

Llama 3.1 released in 8B, 70B, and 405B variants

GPT-4o Mini launched as affordable alternative to GPT-4o

Claude 3.5 Sonnet launched, outperforming Claude 3 Opus

Codestral launched as first dedicated Mistral coding model

GPT-4o launched as first natively multimodal model

GPT-4 Turbo with Vision launched

Claude 3 Haiku launched as fastest Claude model

Claude 3 Opus launched as part of Claude 3 family

Gemini 1.5 Pro launched with revolutionary 1M context window

gpt-3.5-turbo-0125 with improved accuracy

Gemini 1.0 launched in Ultra, Pro, and Nano variants

Llama 2 released in 7B, 13B, and 70B variants

GPT-3.5 Turbo launched