generate-names: default to the frontier models, not the cheap tier

[ralyodio]

Defaults were gpt-4.1-mini and claude-haiku-4-5. Now gpt-5.6-sol and claude-fable-5.

Why

The names showed the cheap tier: generic startup vocabulary, the same few stems recycled, and a drift off-brief on any description longer than a sentence.

And the saving was never real. This tool makes exactly one call per run whatever --count says — that is the whole design, since asking a model for a thousand names directly repeats itself within a few hundred. The model is a rounding error against the value of a name you actually ship, so paying for the weaker one bought nothing.

What the Anthropic path needed first

• max_tokens 4096 → 16000. Fable 5 always thinks, and thinking counts against max_tokens, so a budget that was ample for a non-thinking model can be spent before the JSON starts.
• stop_reason: "refusal" is now reported as a refusal. It arrives as HTTP 200 with no text block, so unhandled it reads as "the model returned nothing" — which sends you to the parser rather than to the answer the API actually gave.

Reading the text block by type rather than by position already handled the leading thinking block; there is now a test pinning that, plus one pinning that this request never grows temperature/top_p/top_k — all three were removed on Fable 5 / Opus 5 / Sonnet 5 and are rejected with a 400.

Heads-up for anyone hitting this

Every Anthropic key in the team vaults is spend-capped until 2026-09-01 (400 invalid_request_error, "You will regain access on 2026-09-01"). So --provider anthropic fails until then and OpenAI is the working path. Verified against a live call today.

Checks

187 tests pass (was 180), typecheck clean.

Verified end to end: the full generate-names … | domainfree pipeline returned 41 available of 100 checked, and a bare run with no --model now reports openai/gpt-5.6-sol.

🤖 Generated with Claude Code

[github-actions]

ThreatCrush Security Scan

6 finding(s)

MEDIUM: 1 | LOW: 5

Severity	Rule	Location
MEDIUM	redos-nested-quantifier	src/domain-free.ts:56
LOW	insecure-temp-file	test/blog.test.ts:73
LOW	insecure-temp-file	test/blog.test.ts:74
LOW	insecure-temp-file	test/credentials.test.ts:43
LOW	insecure-temp-file	test/credentials.test.ts:44
LOW	secret-generic-api-key	test/credentials.test.ts:208

Snippets are redacted; ThreatCrush never prints matched credential material.