#model-migration

Gemini 4 Argon is now GA in the Gemini API at $2/$10 until November 30. What changes when you switch: output caps, caching, sampling and timeouts.

Vendor docs show 10 of 15 current AI models reject a custom temperature or say to drop it. Each rule, where a catalog disagrees, and what to use instead.

GLM-5.3's direct token-metered API went live on August 18. Thinking can no longer be disabled, and the effort tiers are low, high and max only.

Every vendor names its reasoning-effort settings differently and the rungs do not line up. What the ladders look like, and where thinking cannot be disabled.

A same-day LLM eval harness needs 20-50 real tasks, automated grading, and a baseline model to diff against — not hundreds of labels. Qualify a new model fast.

Migrate DeepSeek V3.2 to V4 across open-weight stacks — three reasoning modes, tokenizer change, HCA/CSA attention deltas, KV-cache reduction.

Migrate Gemini 3 Pro to 3.1 with Deep Think — availability tier shift, pricing changes, safety-filter deltas, tool-use schema diff, and rollout plan.

Migrate GPT-5.2 to 5.5 — reasoning-effort behavior change, tool-call schema diff, cost-curve impact, structured outputs, and a phased rollout plan.

Migrate Claude Opus 4.6 to 4.7 — prompt-cache key changes, tool-use schema deltas, 1M context economics, compaction API surface, and rollout plan.

OpenAI retires GPT-4o, GPT-4.1, and o4-mini from ChatGPT on February 13, 2026. Complete migration guide to GPT-5.2 for businesses and developers.