# Starveri API > OpenAI-compatible GPT, Codex, and dedicated Gemma APIs with prepaid credit billing. ## Canonical resources - HTML documentation: https://docs.starveri.net/ - Markdown documentation: https://docs.starveri.net/index.md - Billing dashboard: https://api.starveri.net/billing - API base URL: https://api.starveri.net - OpenAPI schema: https://api.starveri.net/openapi.json - API directory: https://api.starveri.net/apis.json - Model metadata: https://api.starveri.net/models - Support and enterprise: support@starveri.net - Sitemap: https://docs.starveri.net/sitemap.xml ## API surfaces - `POST /v1/chat/completions` — OpenAI-compatible chat completions and SSE streaming. - `POST /v1/responses` — OpenAI-compatible Responses requests and streaming events. - `POST /v1/images/generations` — moderated image generation on the GPT-5.6 Luna backend; accepts a text prompt, n=1..10, and protected reference file IDs. Billing is 0.5 credits per successful output image plus prompt and reference image inputs at GPT-5.6 Luna input pricing; input cost varies with the request, so totals are not fixed. - `GET /v1/models` — authenticated model catalog and pricing metadata. - `POST /v1/files` — file uploads for supported models. ## Model order 1. `gemma-4` 2. `gemma-4-uncensored` 3. `gpt-6-sol` 4. `gpt-6-astra` 5. `gpt-6.1-terra` 6. `gpt-6-luna` 7. `gpt-5.6-sol` 8. `gpt-5.6-terra` 9. `gpt-5.6-luna` 10. `gpt-daybreak-blue` 11. `gpt-5.5` ### Gemma - `gemma-4` — Dedicated serving runtime, 124,160-token context, Standard-only, 5 input / 15 output credits per 1M tokens, automatic request/response safety moderation. - `gemma-4-uncensored` — Dedicated serving runtime, 32,000-token context, Standard-only, 10 input / 30 output credits per 1M tokens, no automatic policy moderation. - Gemma has no non-Standard inference and no input-token caching or cached-input discount. - Both Gemma models expose a binary `thinking` control (`"on"` or `"off"`, default `"on"`). They expose no reasoning modes and do not accept or send reasoning fields; the Playground presents the thinking toggle instead. - During capacity pressure, requests may wait in arrival order. If the dedicated serving runtime is unavailable, requests return HTTP 503 rather than silently substituting another model. `gpt-daybreak-blue` is Standard-only, priced at 3× GPT-5.6 Sol Standard, disabled by default, and approval-gated. Request access from support@starveri.net. Additional safety review and availability controls may apply. ### GPT-6 access and pricing - `gpt-6.1-terra` — 1,050,000-token context; maximum input 922,000; maximum output 128,000. Standard: 50 input / 5 cached input / 62.5 cache write / 150 output credits per 1M tokens. Pro Standard: 150 / 15 / 187.5 / 450. It accepts image input and supports reasoning `none` through `max`. - `gpt-6-luna` — 872,000-token context; maximum input 744,000; maximum output 128,000. Standard: 5 input / 0.5 cached input / 6.25 cache write / 15 output credits per 1M tokens. Pro: 15 / 1.5 / 18.75 / 45. - `gpt-6-sol` — 872,000-token context; maximum input 744,000; maximum output 128,000. Standard: 25 input / 2.5 cached input / 31.25 cache write / 75 output credits per 1M tokens. Pro: 75 / 7.5 / 93.75 / 225. - GPT-6 Luna is the Free-tier default. GPT-6 Sol and GPT-6.1 Terra are available to paid Tiers 1–5 on Standard and Pro, subject to account policy; bans, custom restrictions, rate limits, concurrency limits, and available credits still apply. - GPT-6 Luna and Sol accept text and image input and support reasoning efforts `low`, `medium`, `high`, `xhigh`, and `max`; `none` is not supported. Input above 261,600 tokens uses the 2× long-context model-token rate. Neither model has a Flex profile. ## Context and billing Standard models expose their maximum supported context. Input above 30% of a model's context window is billed at 2× model-token rates. Gemma is Standard-only, so only the long-context multiplier applies. Tool charges are separate. Flex pricing is listed separately where supported. ## Flex Flex is available only for GPT-5.6 Sol and GPT-5.5, always enables web search, and does not preserve context or conversations. GPT-6.1 Terra, GPT-6 Luna, and GPT-6 Sol are not available on Flex. ## Authentication Send `Authorization: Bearer $STARVERI_API_KEY` over HTTPS. The dashboard is at https://api.starveri.net. ### Safe GPT-6 requests ```bash curl https://api.starveri.net/v1/chat/completions \ -H "Authorization: Bearer $STARVERI_API_KEY" \ -H "Content-Type: application/json" \ -d '{"model":"gpt-6-luna","messages":[{"role":"user","content":"Hello"}],"reasoning":{"effort":"high"}}' ``` ```bash curl https://api.starveri.net/v1/responses \ -H "Authorization: Bearer $STARVERI_API_KEY" \ -H "Content-Type: application/json" \ -d '{"model":"gpt-6-sol","input":"Summarize this safely.","reasoning":{"effort":"max"},"max_output_tokens":512}' ```