Moving to GPT-6.1 Sol on TopxAI
· openai, models, api, topxai
The new model ID, lower cache-read prices, reasoning settings and a Responses API example for the GPT-6.1 Sol upgrade.
GPT-6.1 Sol replaces GPT-6 Sol on TopxAI's OpenAI shared and official routes. The model ID is gpt-6.1-sol; the previous ID is retired and does not forward. If an API key has an allowed-model list, update that list to include the new ID.
Update your client
Change gpt-6-sol to gpt-6.1-sol in your SDK, saved client model and Codex configuration. Keep the same TopxAI base URL. Unrestricted keys and route keys without a model restriction can keep using the same key; update any allowed-model list separately. The Codex installer selects the new model by default.
curl https://ai.topxea.com/v1/responses \
-H "Authorization: Bearer $TOPXAI_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"gpt-6.1-sol","input":"Explain this migration in one sentence.","reasoning":{"effort":"medium"}}'Responses is the upstream API for reasoning and tools. TopxAI also accepts Chat Completions and Anthropic-compatible requests and converts them to Responses while retaining the client's response format and streaming tool calls.
Prices and long context
Default USD rates per million tokens, checked against OpenAI's model page on September 30, 2026:
| Category | Shared | Official | OpenAI Standard |
|---|---|---|---|
| Input | $1 | $1.80 | $2 |
| Output | $5 | $9 | $10 |
| Cache read | $0.05 | $0.09 | $0.10 |
| Cache write | $1.25 | $2.25 | $2.50 |
Cache-read rates halve compared with GPT-6 Sol. Input, output and cache-write rates stay the same. The shared route remains 50% of Standard list price; the official route remains 90%.
Once input context exceeds 272,000 tokens, including cached input, the entire request uses the long-context tier: input and cache rates double, output rates multiply by 1.5. The model page shows both tiers. Existing administrator route-price overrides remain in place; accepted requests and historical charges keep their captured prices. A cache miss is not automatically a cache write: billing uses the categories reported by the upstream.
Reasoning settings
The model retains a 1,050,000-token context window and a maximum output of 128,000 tokens. Choose low, medium (the default), high, xhigh or max. GPT-6.1 Sol cannot disable reasoning: TopxAI maps legacy none and minimal, including an Anthropic-style disabled-thinking setting, to low.
For Codex, set model = "gpt-6.1-sol" in ~/.codex/config.toml; model_reasoning_effort = "medium" is a supported starting point. The Codex guide covers the full setup.
Models: gpt-6.1-sol · Models & pricing
Related
- GPT-6.1 Sol API pricing
- Codex CLI with TopxAI: download, install and connect
- How requests are billed
- Moving to Claude Sonnet 5.5 on TopxAI
- TypeSafe Jev: a System One model that returns judgments, not text
Other languages: 简体中文