GPT-5.5 — OpenAI doubled the GPT-5 line, local hardware just got cheaper by comparison

GPT-5.5 launched at $5 input / $30 output per 1M, exactly 2× the GPT-5.4 rates. OpenAI claims ~40% fewer output tokens on Codex tasks, which mostly offsets the increase to a ~20% net rise. Crucially, that ~20% comes out of cloud users' pockets; local-hardware buyers get a free improvement to their break-even math.

Verdict: Cloud doubled the per-token price; local payback math shifted

The take

OpenAI shipped GPT-5.5 on April 23, available via Chat Completions and Responses APIs at $5.00 in / $30.00 out per 1M with a 1M context window. GPT-5.5-pro lands at $30 / $180. Both API endpoints went live April 24.

The framing OpenAI led with — '40% fewer output tokens on Codex' — is real but partial. For Codex-style coding workloads it nets out to roughly +20% effective cost vs GPT-5.4. For non-Codex workloads (chat, generic reasoning, document summarization) it's the full 2× hit, no token-economy offset. Don't take the headline number as universal.

What this changes for local: the cost calculator on this site already lists GPT-5.5. Rerun it with your actual usage and you'll see the break-even months drop. A heavy user (50M tokens/month) running against GPT-5.4 sat at ~$438/month cloud, ~5 months payback on a $2,000 setup. Same usage on GPT-5.5: ~$875/month cloud, ~2.5 months payback. Local hardware costs are unchanged; cloud just got more expensive, so the ratio flipped.

Practical takeaway: if you've been running GPT-5.4 heavily and considering local, the spreadsheet just got materially friendlier. If you've been running ChatGPT Plus and rarely hit limits, this changes nothing for you. The flat-rate plans (Plus / Pro / Max 5x / Max 20x) are unaffected; only the per-token API tier moved.

Where this fits

Hardware: NVIDIA RTX 5090 · Mac Studio M4 Max 64 GB · NVIDIA RTX 4090 · AMD Radeon RX 9070 XT

Sources

Next step

Try this in the planner→