Changelog
LAST UPDATED
AUG 2026
Pricing, model, and feature changes for FreeAI.DevTools, most recent first.
§ 2026-08-02
NEW MODELPRICINGFEATURE
- Added 9 new models: Claude Fable 5, Claude Opus 5, Claude Sonnet 5 (intro pricing $2/$10 through Aug 31), GPT-5.6 Sol, Terra & Luna, Grok 4.5, Gemini 3.6 Flash, Gemini 3.5 Flash Lite. We now track 52 models.
- Grok 4.20 price cut applied: input $2.00 → $1.25, output $6.00 → $2.50 (verified against xAI).
- DeepSeek V4 Flash updated to the official 0731 build: $0.09/$0.18 per 1M tokens, released July 31.
- Applied OpenRouter drift for DeepSeek V3.2, Mistral Small 3.1, Llama 4 Maverick, Llama 3.3 70B; DeepSeek context windows corrected to 163,840.
- GPT-5.6 Terra and Luna list prices taken from OpenAI docs ($2/$12 and $0.20/$1.20 after the July 30 cut); OpenRouter serves discounted routing rates for both.
- Model counts across the site now derive from the dataset itself, so they can no longer drift.
- New pricing-history archive: every sync stores a dated snapshot of the full dataset.
Prices verified against provider pricing pages (OpenAI developer docs, Anthropic, xAI) with OpenRouter as the drift detector.
§ 2026-04-26
NEW MODELPRICING
- Added 6 new models: GPT-5.5, GPT-5.5 Pro (1M context, premium tier), DeepSeek V4 Pro, DeepSeek V4 Flash (1M context, large drop in cost), Grok 4.20 (2M context), Mistral Small 4.
- Bumped DeepSeek V3.2 pricing: input $0.28 → $0.252, output $0.42 → $0.378 (10% drop).
- Expanded Claude Sonnet 4 + Sonnet 4.5 context windows from 200K to 1M tokens.
- Corrected GPT-4.1 family context to 1,047,576 (matches OpenAI documentation exactly).
- Marketing copy updated: 35+ → 40+ models site-wide (we now track 43 models across 7 providers).
Auto-applied from npm run sync:models -- --apply against OpenRouter; new models curated by hand.
§ 2026-04-24
AUDITPRICINGNEW MODEL
- Verified all 36 models against live provider pricing pages.
- Added Claude Opus 4.7 (Apr 2026 release): $5/$25 per 1M tokens, 1M context, Frontier reasoning.
- Updated 6 model prices: Gemini 2.5 Flash, DeepSeek R1, Mistral Large 3, Mistral Small 3.1, Llama 4 Maverick, Llama 3.3 70B.
- Corrected 18 context-window values to match provider documentation (e.g., GPT-5 family bumped to 400K, Claude Sonnet 4.6 expanded to 1M).
Verified against OpenRouter + each provider's official pricing page.