AI Studio is still free for adults in supported countries, but since April 20, 2026, Gemini Pro and Nano Banana need Google AI Pro/Ultra or a paid API key.
Plus is $20 a month and the API is billed separately. Short, occasional chats cost less on the API; long daily threads, reasoning and Codex favor Plus.
No provider wins for every model. As of September 26, 2026, DeepSeek direct off-peak costs half of hosts charging its peak rate; check versions and router fees.
As of September 26, 2026, GPT-6 Luna and DeepSeek Flash cost least per token. Among $2-input models, Grok 4.7 is cheapest; past 272K tokens, Sonnet 5 beats Sol.
Same Claude models, different jobs: the API is per-token access for your own code; Claude Code is a coding agent that runs on Pro/Max or an API key.
As of September 26, 2026, GPT-6 Sol matches Sonnet 5 and Astra matches Fable 5.1 on list price. Prompts over 272K favor Claude; Luna costs a tenth of Haiku 4.5.
DeepSeek doubles API rates from 01:00–04:00 and 06:00–10:00 UTC on weekdays. Flash was cut on September 10, 2026, so waiting saves less per batch.
Start with Gemini's free tier unless your prompts are private or you're outside its regions; add Groq or OpenRouter for headroom. As of September 26, 2026.
As of September 2026, the free Gemini app runs Nano Banana 2 with no fixed image count, AI Mode allows 25 images a day, and every Gemini API image is billed.
As of September 2026, Google prints no image count for the free Gemini app: usage refreshes every 5 hours up to a weekly cap. AI Mode allows 25 images a day.
In the official Python and Node SDKs, base_url beats OPENAI_BASE_URL and endpoint paths are appended to it, so paste the base ending in /v1, not an endpoint.
As of September 26, 2026, grok-4.7 costs $2 input and $6 output per million tokens; grok-4.3 costs $1.25 and $2.50. No free API allowance is listed.
A Claude 400 on Amazon Bedrock can be a stripped beta header, old Claude Code, full context, a bad model ID or an account setting. Fix it by the exact message.
Claude Fable 5.1 costs $10 per million input and $50 per million output tokens. Only cache reads got cheaper, $1 to $0.25, so savings track your cache use.
GPT-6 Sol is the cheaper default for short-context work. Opus 5.5 scores higher, and its 2x price gap shrinks with cache reads and can flip above 272K input.
Compare GPT Image 2.5 Sunburst and Flare token costs with LaoZhang’s $0.03/call VIP option, including quality settings, editing charges, and monthly budgets.
GPT Image 2.5 runs in ComfyUI 0.35.0+ as a Partner Node billed in Comfy credits, not an OpenAI key. A 1024×1024 image costs about $0.0084 at low quality.
n8n's OpenAI node runs GPT Image 2.5 up to high quality and 1536 px but drops token usage. For xhigh, 4K, transparency or URL replies, use HTTP Request.
ChatGPT Images 2.5 and the API have no denoise setting. Find the first file with the grain, restart from the last clean one, and change one thing per retry.
Start with Flare for drafts and volume; use Sunburst when an edit must keep a product or layout intact. Same cost per attempt; tests put Flare 25–45% faster.
The Free tier's $100/month is a usage ceiling, not credit. Only a credit grant in your account or eligible data-sharing daily tokens make requests free.
ModelScope runs a free Qwen-Image-2.1 API once you bind a real-name-verified Alibaba Cloud account. At 1 credit per call, that tops out near 250 calls a day.
Use a one-color tool for flat logos, a background remover for photos, and ChatGPT for new art. Save as PNG, then test it on a dark background before sending.
The cloud credit is Anthropic's one-time $100/$250 trial for cloud sessions, usage credits bill your card at API rates, and a reset refills one limit for free.