Changelog

Recent changes

1. 2026-09-15 models

   Fourteen models added after a full re-verification against every provider's official pricing page. Anthropic: Claude Opus 5 ($5/$25), Claude Sonnet 5 ($2/$10), and Claude Fable 5.1 ($10/$50). Sonnet 5 is notable twice over, it is the cheapest frontier Claude yet and the increase to $3/$15 that was scheduled for 1 September 2026 was cancelled, so the launch price is now the standard price. Fable 5.1 prices cache hits at 0.025x base input instead of the usual 0.1x, which is $0.25 per 1M. OpenAI: GPT-6 Astra ($10/$50), GPT-5.6 Sol ($4/$20), GPT-5.6 Terra ($2/$12), and GPT-5.6 Luna ($0.20/$1.20). Google: Gemini 3.8 Flash ($0.75/$3.75 promotional through 31 December 2026, then $1.50/$7.50) and Gemini 3.5 Flash-Lite ($0.30/$2.50). xAI joins as a new provider with Grok 4.6 ($2/$6), Grok 4.5 ($2/$6), and Grok 4.3 ($1.25/$2.50), all three tiered so a prompt that reaches 200k tokens is billed at the higher rate for every token in the request. Groq: GPT OSS 120B ($0.15/$0.60) and GPT OSS 20B ($0.075/$0.30).

2. 2026-09-15 pricing

   Two status corrections. Groq moved Llama 3.3 70B Versatile and Llama 3.1 8B Instant to Enterprise "contact sales" pricing on 26 August 2026 and retired its public pricing page on 11 August 2026, so Llama 3.3 70B and Llama 4 Scout are now flagged "delisted": their pages stay up with the last published rates for reference and they are out of every calculator. Going the other way, GPT-4o, GPT-4o mini, o3, and o3-mini are back on OpenAI's public price list, so they return to "official" and rejoin the calculators at $2.50/$10, $0.15/$0.60, $2/$8, and $1.10/$4.40. Every remaining row was re-checked against its provider page and all prices were unchanged. Claude Code plan prices re-verified, still $20, $100, and $200.

3. 2026-06-13 feature

   The verdict tool now reads your whole ccusage history, not just the latest month. A month-by-month table shows API-equivalent value per month, splitting interactive Claude Code usage (what a subscription covers) from automation and non-Claude API calls, with every multiple tracking the plan you select. A year-in-review receipt card sums it into one shareable image with a per-month trend. The headline verdict uses your latest complete month so a partial current month does not understate it.

4. 2026-06-13 feature

   Open data rails: the full pricing dataset is now free at /api/pricing.json (regenerated from the same source the site renders, with per-row last-verified dates), plus llms.txt and llms-full.txt for AI assistants. Parser hardening from real-world pastes: terminal noise around the JSON is stripped, ccusage "period" month keys are read correctly, and dot-notation model ids like gpt-5.5 now match verified price rows.

5. 2026-06-12 feature

   The Claude Code page is now a verdict tool: paste one month of ccusage output and get the API-equivalent bill recomputed from raw tokens at our verified prices, a value multiple against your plan, a break-even table, and a shareable receipt card you can copy or download as an image. Everything is parsed in your browser; usage data is never uploaded. The preset estimator remains below for anyone without usage data.

6. 2026-06-12 models

   Ten models added, all verified against official provider pages: Claude Fable 5 ($10/$50, cache read $1) and Claude Opus 4.8 ($5/$25) from Anthropic; GPT-5.5 ($5/$30), GPT-5.4 ($2.50/$15), GPT-5.4 mini ($0.75/$4.50), and GPT-5.4 nano ($0.20/$1.25) from OpenAI; Gemini 3.1 Pro Preview ($2/$12, long-context tier $4/$18), Gemini 3.5 Flash ($1.50/$9), and Gemini 3.1 Flash-Lite ($0.25/$1.50) from Google; Llama 4 Scout ($0.11/$0.34) hosted by Groq.

7. 2026-06-12 pricing

   GPT-4o, GPT-4o mini, o3, and o3-mini no longer appear on OpenAI's public pricing page. Their rows are flagged "delisted": pages stay up with the last verified rates for reference, and they are removed from all calculators. Corrections to existing rows: Anthropic max output figures fixed (Opus 4.7 is 128k, Sonnet 4.6 and Haiku 4.5 are 64k), Haiku 4.5 extended thinking flag corrected, and a tokenizer caveat added for Opus 4.7 and later (roughly 30% more tokens for the same text than older Claude models). Claude Code plan prices re-verified, unchanged.

8. 2026-05-06 pricing

   Full verification pass complete: all 10 model rows now flagged "official". Anthropic (via platform.claude.com): Opus 4.7 refreshed to the current Opus 4.5+ tier ($5/$25 per 1M); Sonnet 4.6 and Haiku 4.5 confirmed. OpenAI (via platform.openai.com): o3 refreshed to its mid-2025 reprice ($2/$8, was $15/$60 at launch); 4o, 4o-mini, o3-mini confirmed. Google (via ai.google.dev): Gemini 2.5 Pro cache pricing corrected to $0.125 / $0.25 (≤200k / >200k tier); 2.5 Flash cache pricing corrected to $0.03. Groq (via console.groq.com): Llama 3.3 70B Versatile relabeled with provider "groq" (model_family meta-llama) since Meta does not host this model directly; $0.59 / $0.79 input/output confirmed against Groq's console.

9. 2026-05-06 plans

   Verified Claude Code plan prices against claude.com/pricing: Pro $20/mo (or $17/mo billed annually at $200), Max from $100/mo, Max 20x at $200/mo. Plan data flipped to "official".

10. 2026-04-28 launch

    Initial pricing data seeded for 10 models across Anthropic, OpenAI, Google, and Meta. All entries marked as inferred. Verification pass against official provider docs is the next priority before any production use of these numbers.