Move autoblog to the current frontier models on both providers - #229
Merged
Merged
Conversation
Article and guest-post text now run on claude-opus-5 (was claude-opus-4-8) and gpt-5.6-sol (was gpt-5.5). Both are env-overridable — BACKEND_AI_ANTHROPIC_MODEL is new, BACKEND_AI_OPENAI_MODEL already existed — so the next release or price change is a variable, not a deploy. generateStructuredOutput sent `thinking: disabled` to every Anthropic model. On Opus 5 that makes the model occasionally write reasoning into the visible answer, and Fable 5.1 rejects it outright. Thinking is now adaptive on 4.6+ models and stays disabled on Haiku 4.5, which predates adaptive. The article max_tokens rises 48k -> 64k because reasoning tokens share that ceiling. The spend ledger learns gpt-5.5 and the three gpt-5.6 tiers; every 5.x id starts with "gpt-5", so without exact rows the prefix match billed them at the 2025 gpt-5 rate. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01AFoGhPD5bGsQ5StFqrF6ck
ThreatCrush Security Scan39 finding(s) HIGH/CRITICAL: 2 | MEDIUM: 28 | LOW: 9
Snippets are redacted; ThreatCrush never prints matched credential material. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Autoblog article and guest-post text now run on the current top-tier model of each provider:
claude-opus-4-8claude-opus-5BACKEND_AI_ANTHROPIC_MODEL(new)gpt-5.5gpt-5.6-solBACKEND_AI_OPENAI_MODELBoth are env-overridable so the next release or price change is a Railway variable, not a deploy.
Why the thinking change
generateStructuredOutputsentthinking: { type: "disabled" }to every Anthropic model. On Opus 5 that occasionally makes the model write its reasoning into the visible answer, and Fable 5.1 rejects it with a 400. Thinking is now adaptive on 4.6+ models and stays disabled on Haiku 4.5 (which predates adaptive and only knowsbudget_tokens). The articlemax_tokensgoes 48k -> 64k because reasoning tokens come out of the same ceiling; the call already streams so the larger cap has no timeout cost.Spend ledger
rateFor()matches on the longest known prefix, and every 5.x id starts withgpt-5, sogpt-5.5andgpt-5.6-*were being billed at the 2025gpt-5rate. Added rows forgpt-5.5($5/$30) and the threegpt-5.6tiers (Sol at the $4/$20 promotional rate published through 2026-11-21).Verified
tsc --noEmitclean, full vitest suite green locally (152 files, 1997 tests).ANTHROPIC_API_KEYis capped (regain access on 2026-10-01) and the RailwayOPENAI_API_KEYreportsno credits remaining. Autoblog is not publishing at all right now regardless of this PR; both accounts need topping up.Not touched
Haiku 4.5 callsites (keyword research, blog detection, pitches), the ads creative generator, apply-fix, and the audit engines keep their pins. Haiku 4.5 is still the current Haiku.
🤖 Generated with Claude Code
https://claude.ai/code/session_01AFoGhPD5bGsQ5StFqrF6ck