Anthropic and OpenAI both cut prices on the same Tuesday, and Anthropic's cut arrives attached to three migration errors.

• GPT-5.6 Sol $4 / $20 → GPT-6 Sol $2 / $10 per MTok.
• GPT-5.6 Luna $0.20 / $1.20 → GPT-6 Luna $0.10 / $0.50 per MTok.
• Claude Opus 5 $5 / $25 → Claude Opus 5.5 $4 / $20 per MTok.
• All six figures are from the vendors' own pages, read 23 September. The GPT-5.6 numbers are described by OpenAI as promotional pricing, so the comparison is against a promotion, not a list price.

{
  "prompt_cache_diagnostics": {
    "type": "cache_miss",
    "reason": "tools_changed",
    "comparison_reusable_tokens": 5629,
    "cache_missed_tokens": 5629
  }
}

A Prompt Caching Dashboard reports hit rates over time and splits input into cached and uncached tokens.

OpenAI quotes GitHub's Mario Rodriguez saying Copilot cut the share of prompt tokens needing fresh processing by more than 50% across billions of requests against their previous baseline. That is a customer testimonial on a vendor page, not a measurement we can reproduce, and we print it as such.³

Prewarming lets an application process known context — shared instructions, tool definitions, reference material — before the first user request arrives.

OpenSSH 7.2p1 · JSch 0.1.66 (from GitHub's named fork) · TeamCity 2021.2.3 · Go SSH 0.16.0 · libssh2 1.11.0 · PuTTY 0.82.

GitHub's note distinguishes the key type ssh-rsa, which covers all RSA keys, from the signature type ssh-rsa, which means SHA-1 specifically. Only the signature type is being removed.

GitHub Enterprise Server picks the removals up in 3.25, and the ML-KEM addition in 3.24. The ML-KEM key exchange does not come to the U.S. region of GitHub Enterprise Cloud with Data Residency.⁴

One date in GitHub's own schedule reads "January 13, 2026", which is in the past; from the sequence it is plainly January 2027. We have not seen a correction on the page.